Neuropsychiatric Traits Associated with Refractory Impulse Control Disorder in Parkinson’s Disease

Impulse control disorder (ICD) is common and an important nonmotor symptom in Parkinson’s disease (PD), which is strongly associated with dopamine agonist therapy [1, 2]. Thereby, discontinuation of dopamine agonist has been recognized as a first-line therapeutic option to control ICD in PD [1, 2]. However, in certain PD patients, the interventional therapy does not improve ICD behaviors [3], and individual neuropsychiatric traits have been suggested to contribute to ICD in PD [2].

Given the individual variations, we hypothesized that specific neuropsychiatric traits may have predictive value for persistent or refractory ICD after dopamine agonist substitution therapy, and therefore we conducted a case-control study using a multicenter trial database (REIN-PD trial) [4].

In the REIN-PD trial (NCT01683253), behavioral and neuropsychiatric characteristics of PD patients with ICD, were assessed before and 12 weeks after the intervention of substituting dopamine agonists with equivalent-dose levodopa [4]. All participating subjects had been stably treated with dopaminergic drugs for >6 months, and patients were excluded when they had dementia, a premorbid history of ICD, or a history of psychiatric illness [4].

Thirty-eight out of 50 PD-ICD patients in the REIN-PD trial completed the first 4 weeks of the substitution period. Among them, 2 developed dopamine agonist withdrawal syndrome, and an additional 2 showed parkinsonism aggravation during the 8-week maintenance period and were dropped out before the last visit at 12 weeks. The poor-outcome group was defined as patients with a paradoxical increase in any of the four subitem scores of the modified Minnesota Impulsive Disorders Interview (mMIDI) (compulsive shopping, compulsive gambling, hypersexuality, or compulsive eating) at the endpoint assessment and included those 2 dopamine agonist withdrawal syndrome and 2 parkinsonism aggravation cases who were assessed with mMIDI at the time of their dropout. We opted for this definition because MIDI has been used in previous clinical studies on ICD in PD [5, 6], and it has shown a good test-retest reliability [7].

We assigned subjects to the good-outcome group if they met the following four criteria: (1) no increase in any of the four sub-item scores of mMIDI; (2) reduction in the total mMIDI score; (3) absence of new abnormal behavior after the intervention; and (4) no significant adverse events during the trial.

We collected PD-associated clinical information including dopaminergic drug dosage, the Unified PD Rating Scale (UPDRS), the Hoehn and Yahr (H&Y) stage, and the Mini-Mental State Examination (MMSE) score. We also obtained pre-interventional scores of the Korean Beck Anxiety Inventory (K-BAI) [8], Korean Beck Depression Inventory (K-BDI) [9], Korean Barratt Impulsiveness Scale-11 (K-BIS-11) [10], Korean Obsessive-Compulsive Inventory-Revised (K-OCI-R) [11], Korean State-Trait Anger Expression Inventory (K-STAXI) [12], and Korean Neuropsychiatric Inventory (K-NPI) [13] assessed at baseline in the REIN-PD participants.

Changes in mMIDI item scores between the groups were compared using the Wilcoxon signed-rank test, and clinical variables were compared using the Mann-Whitney or Fisher’s Exact test. We also evaluated the predictability of a trait for the poor outcome using the receiver-operating characteristic (ROC) curve analysis and determined the cutoff scores using Youden’s index [ = sensitivity – (1–specificity)] [14]. SPSS version 26.0 was used with a significance level set at 0.05 (two-tailed).

We identified 15 (39.5%) patients in the poor-outcome group and 20 patients in the good-outcome group. Three subjects were undetermined because they exhibited compulsive repetitive behaviors newly after the intervention despite improvements in all four ICD behaviors (Fig. 1).

In comparison to the poor-outcome group vs. good-outcome or intention-to-treat total-ICD (N = 50) groups, there were no differences in age, gender, disease duration, levodopa equivalent dose, Unified PD Rating Scale (UPDRS) scores, Hoehn and Yahr stage (H&Y), and global cognition score (Table 1).

Baseline for ICD scores as in each subitem of the -mMIDI did not significantly differ between the poor- and good-outcome groups (online suppl. Table 1; see www.karger.com/doi/10.1159/000507447 for all online suppl. material). After 12 weeks of the intervention, all four subitem scores of mMIDI significantly improved in the good-outcome group, whereas there were no changes or slight increases in mMIDI scores in the poor-outcome group (Fig. 2a). The proportion of ICD subtypes at baseline did not differ between the two groups, and there were total 18 cases of paradoxical increases in mMIDI after the intervention (Fig. 2b).

The poor-outcome group showed higher levels of obsessive compulsiveness, anger expression, and anxiety compared with the good-outcome group at baseline (p = 0.007 for Korean State-Trait Anger Expression Inventory [K-STAXI]; 0.009 for Korean Obsessive-Compulsive Inventory-Revised [K-OCI-R]; 0.043 for Korean Beck Anxiety Inventory [K-BAI]; Table 1). The degree of obsessive compulsiveness of the poor-outcome group was also higher when comparing to the intention-to-treat total-ICD group (N = 50) and to the subjects excluding the poor-outcome group (n = 35) (p = 0.054 and 0.009 for K-OCI-R; Table 1, online suppl. Table 2).

In the ROC curve analysis for discriminating the patients with poor outcome among the intention-to-treat total-ICD population (N = 50), the K-OCI-R showed the highest area under the curve (AUC) as 0.735, with 80.0% sensitivity and 74.3% specificity, and the optimal cutoff of the K-OCI-R score was 22/23 (Fig. 2c, online suppl. Table 3). While, AUC of the K-STAXI was at the suboptimal level as 0.659, with 54.3% sensitivity and 80.0% specificity.

This study revealed that PD-ICD patients with pre-existing obsessive compulsiveness were likely to exhibit paradoxical aggravation of ICD behaviors after dopamine-agonist switch-off. Although the poor-outcome group showed borderline high LED compared to the good-outcome group at baseline, LED after the intervention was not significantly different between the two groups (1,100 ± 640 vs. 915 ± 484, p = 0.483).

ICD in PD, characterized by repetition of rather simple behaviors, may be associated with the finding of a high incidence of obsessive compulsiveness in PD patients with ICD [2]. Because ICD is a behavioral addiction, it can be postulated that a high degree of obsessive compulsiveness may be related to an activity shift from the ventral to the dorsal part of the basal ganglia with habituation of addictive behaviors [15]. With this habituation, addictive behaviors can become more compulsive and be difficult to treat by medication change alone. In neuroimaging studies in PD with ICD patients, deactivation of the inhibitory network has shown to be initiated by the orbitofrontal cortex (OFC) and anterior cingulate area in response to dopamine agonist administration [16, 17]. Interestingly, aberrant OFC-ventral striatal circuit and a reduced OFC volume are related to obsessive-compulsive disorder [18, 19], similar to OFC dysfunction in PD patients with ICD.

There is also an issue of co-existing stereotypical repetitive behaviors, known as ‘punding’ in PD patients with ICD [20]. In the REIN-PD trial, the degree of improvement of punding following dopamine agonist discontinuation was relatively small (56.2% improvement) in contrast to 77.8–90.5% improvement in ICD behaviors [4]. Punding is reported to be weakly associated with dopamine agonist, whereas it is relatively strongly associated with high-dose levodopa [3, 20]. Our data suggests that punding-like behaviors need to be monitored when implementing a dopamine-agonist switch-off therapy in PD patients. Together with these observations, PD patients with chronic levodopa treatment tend to have high impulsivity and risk-taking behaviors as with those taking dopamine agonists, especially when they are taking high doses [21-23]. The poor-outcome group had a high -levodopa dose relative to the good-outcome group at baseline in this trial (daily levodopa dose = 0.831.7 ± 637.2 vs. 550.3 ± 508.3 mg/day, respectively, p = 0.240, and total levodopa daily equivalent dose = 1,436.9 ± 858.0 vs. 978.7 ± 696.1 mg/day, respectively, p = 0.064). Therefore, their decisional impulsivity and risk-taking trait might not be easily overcome by switching dopamine agonists into -levodopa, which results in a further increase in the -levodopa dose.

The aggression/anger traits are related to impulsiveness, and patients with intermittent explosive disorder, a disorder of impulsive aggression, tend to have conco-mitant ICD [24, 25]. In PD, aggression as well as depression, anxiety, alexithymia, and obsessive-compulsive traits have been reported to be risk factors of ICD [2]. In addition to the literature, we demonstrated that high levels of obsessive compulsiveness and anger may be related to more complicated and refractory ICD in PD.

Despite reporting novel clinical insights, the sample size of this study was relatively small. However, the data was prospectively collected in a multicenter trial setting, and the poor versus good outcomes were strictly defined and systematically evaluated in a prospective setting. Therefore, this study would have more credibility than any retrospective studies or clinical trial without assessing ICD as a principal outcome. The mMIDI has not yet been validated for ICD severity assessment in PD. However, we planned this study before the publication of the QUIP-RS, a recommended scale for the severity classification of ICD [26]. Notwithstanding, the use of mMIDI was not perceived to critically affect the current study results because the primary aim of this study was not to assess the efficacy of dopamine agonist switch-off, but to assess neuropsychiatric trait changes by the drug switch-off. Further large-scale studies with validated assessment tools are required to convince our observations in the future.

In conclusion, the present study suggests that pre-existing obsessive compulsiveness may predict poor response in treating ICD in PD, and the validity of OCI-R is worth being replicated by further studies involving various PD populations.

This study was performed in accordance with the Declaration of Helsinki, and all participants of this study have given their written informed consent. The study protocol has been approved by the institutional review board of each participating center.

The authors have no conflicts of interest to declare.

This study was funded by Research Grant of Sandos Korea, Ltd. The funder had no role in the design or preparation of the study or manuscript.

J.-H.C.: design, analysis of the data, drafted the manuscript for intellectual content; J.-Y.L.: design and conceptualization of the study, acquisition of data, analysis and interpretation of the data, drafted the manuscript for intellectual content. B.J., S.-B.K., W.T.Y., H.-W.L., O.D.K., J.W.K., J.-M.K., H.-I.M., H.T.K., J.S.B.: acquisition of data, interpretation of data, critical review of the manuscript. J.W.C.: major role in the acquisition of data, interpretation of the data, revised the manuscript for intellectual content.

Ji-Hyun Choi and Jee-Young Lee contributed equally to this work.