The Association of Antidepressant Use and Impulse Control Disorder in Parkinson’s Disease

Abstract

Objectives:

To examine whether initiation of an antidepressant is associated with the development of impulse control disorder (ICD) in patients with Parkinson’s disease (PD).

Design:

We performed a retrospective analysis utilizing data from the Parkinson’s Progression Markers Initiative (PPMI). Two-sample Mann-Whitney tests were used for comparison of continuous variables and Pearson χ² tests were used for categorical variables. Kaplan-Meier survival analysis and cox proportional hazards regression analysis was used to assess the hazard of ICD with antidepressant exposure.

Setting:

The PPMI is a multi-center observational study of early PD with 52 sites throughout North America, Europe, and Africa.

Participants:

Participants in the current study were those in the PPMI PD cohort with a primary diagnosis of idiopathic PD.

Measurements:

The presence of ICD was captured using the Questionnaire for Impulsive-Compulsive Disorders in Parkinson’s Disease (QUIP). Antidepressant use was defined based on medication logs for each participant. Depressive symptoms were captured using the Geriatric Depression Scale (GDS).

Results:

A total of 1,045 individuals were included in the final analysis. There was a significant increase in the probability of ICD in those exposed to serotonergic antidepressants compared to those not exposed (Log-rank p<0.001). Serotonergic antidepressant use was associated with a hazard ratio for ICD of 1.4 (95% CI 1.0 – 1.8, z-value 2.1, p = 0.04) after adjusting for dopamine agonist use, depression, bupropion use, MAOI-B use, amantadine use, LEDD, disease duration, sex, and age.

Conclusions:

Serotonergic antidepressant use appears to be temporally associated with ICD in patients with PD.

Introduction

Background

Parkinson’s disease (PD) is a disabling neurodegenerative condition with varied psychiatric manifestations that contribute at least as much to disability and decreased quality of life as the core motor symptoms that define the disease [1–4]. One of PD’s most challenging psychiatric syndromes is impulse control disorder (ICD), which impacts 20–45% of patients [5–7]. Impulse control disorders encompass a diverse array of dysregulated behaviors that include excessive gambling, sex, spending, overeating, goal-directed activities (writing, painting, gardening, collecting, etc.), repetitive motor activities (cleaning, examining, sorting, etc.), and walking/driving without purpose [5, 6]. Impulse control disorders in PD have classically been considered iatrogenic phenomena with exposure to dopamine agonists being the primary driver; however, additional risk factors, including depression, male sex, disease duration, and medications like monoamine oxidase B inhibitors (MAOI-B), and amantadine have also been implicated [5, 8]. While progress in elucidating the etiology and common risk factors of ICD in PD continues, it remains a challenging clinical problem without a definitive understanding of the underlying neurobiology.

Overlap with Hypomania

A striking clinical feature of ICD in PD is the overlap with symptoms of hypomania observed in bipolar disorder. Hypomania often presents with similar irrepressible behavior, including excessive spending, risky sexual activities, and frenzied goal-directed activity accompanied by grandiosity and decreased need for sleep [9]. Like the established association between dopamine agonists and ICD, antidepressant use increases the risk of hypomania/mania in susceptible individuals with bipolar disorder [10–19]. Antidepressants have long been recognized as precipitants of abnormal mood elevation in certain patients [20, 21]. Given the significant clinical similarity between ICD and hypomania, the role of antidepressants in contributing to ICD warrants investigation.

Few studies have examined the association between antidepressant use and ICD – the few that have explored this link were exploratory, cross-sectional analyses [22–24]. To our knowledge, no studies have conducted a comprehensive longitudinal examination of the association between antidepressant initiation and the emergence of ICD in PD.

Given this background, the primary aim of our study is to examine whether initiation of an antidepressant is associated with the subsequent development of ICD in patients with PD. We hypothesize that antidepressant use is associated with subsequent ICD even after adjustment for known ICD risk factors. Findings from this study will help guide future work to further elucidate the neurobiology of ICD in PD in support of developing more targeted and personalized therapy for this challenging neuropsychiatric condition.

Methods

Study Design and Data Source

To test our hypothesis, we performed a retrospective analysis utilizing data from the Parkinson’s Progression Markers Initiative (PPMI), a multi-center observational study of early PD (http://www.ppmi-info.org/data). The organization, aims, and methodology of the PPMI has previously been published in detail [25]. The PPMI study focuses recruitment on individuals with early-stage PD (disease duration ≤2 years at study entry) who are not expected to require PD medication for at least 6 months from the baseline visit. Evaluations are conducted on a roughly annual basis and include motor assessments, neurobehavioral/neuropsychiatric testing, autonomic, olfaction, and sleep testing as well as neuroimaging including DAT SPECT and structural MRI. Neuropsychiatric symptoms are derived from standardized questionnaires and not based on formal psychiatric diagnostic criteria. The PPMI includes 52 sites throughout North America, Europe, and Africa. All participating PPMI sites have obtained approval from their respective Institutional Review Boards, and written informed consent is obtained from all participants. Data for the present study were collected between July of 2010 and June of 2023. Data used in the preparation of this article were obtained on July 18th, 2023 from the Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/access-data-specimens/download-data), RRID:SCR_006431. For up-to-date information on the study, visit www.ppmi-info.org.

Primary and Secondary Study Outcomes

Our primary outcome of interest was the hazard of ICD in those using antidepressants before and at the time of ICD measurement compared to those not using antidepressants. Covariates of interest included the association of depression, disease duration, age, sex, levodopa daily equivalent dosage (LEDD), dopamine agonist use, MAOI-B use, and amantadine use on the hazard of ICD.

Outcome and Exposure Measures

The presence of ICD was captured using the Questionnaire for Impulsive-Compulsive Disorders in Parkinson’s Disease (QUIP), a validated questionnaire developed to screen for ICD in PD [26]. The QUIP has been shown to reliably detect ICD in participants with PD with a sensitivity of 0.94 and a specificity of 0.72 [26]. ICD was treated as a binary variable in our analysis and considered as “present” if the total QUIP score exceeded one and “absent” if the total QUIP score was zero. We did not count positive ICD screens at baseline as ICD events in our analysis given the inability to definitively determine when the ICD emerged in relation to antidepressant exposure. Similarly, we did not include individuals with ICD at every visit as ICD events for the purpose of our primary analysis. The QUIP score was calculated by adding one point for a positive response to questions regarding behaviors or excess thinking about any of the following: (1) gambling behavior, (2) sexual behavior, (3) buying behavior, and (4) eating behavior. One point was also added for a positive response to any of three additional questions about (1) excessive time spent on hobbies/organized activities, (2) simple motor activities, and (3) walking/driving without an intended purpose. An additional point was added for excess dopaminergic medication use or experiencing a strong desire for more Parkinson’s medication. This ICD case definition using the QUIP has been validated and used in prior studies of ICD in PD using PPMI [26–28].

Antidepressant, dopamine agonist use, MAOI-B use, and amantadine use were defined based on medication logs for each participant. Antidepressant use was coded as present if a participant was prescribed any selective serotonin reuptake inhibitor (SSRI), serotonin-norepinephrine reuptake inhibitor (SNRI), tricyclic antidepressant (TCA), or atypical antidepressant (mirtazapine, bupropion, buspirone, atomoxetine, vilazodone, or trazodone). Antidepressants were also separated into serotonergic agents and non-serotonergic agents. The only non-serotonergic antidepressant included in the analysis was bupropion, which was analyzed separately in regression models. Exposure to a dopamine agonist was coded as present if a participant was prescribed any of the following: ropinirole, rotigotine, pramipexole, or apomorphine. Participants were coded as using antidepressants or dopamine agonists before developing an ICD if the start date for the medication occurred before the date when a QUIP score was calculated, and the stop date for the medication happened after or at the time the QUIP score was calculated.

Levodopa equivalent daily dosage (LEDD) was calculated for each participant at each study visit and reported as part of the curated PPMI dataset. The LEDD constitutes the total cumulative exposure to all dopaminergic drugs at each visit and was calculated using previously validated conversion methods [29].

Depressive symptoms were captured using the Geriatric Depression Scale (GDS), a 15-item measure that has been well-validated as a depression screener in patients with PD [30, 31]. The GDS was used as a binary variable in regression models, with a score greater than or equal to 5 (mild depression) constituting the presence of depression and a score less than 5 comprising no depression. The cut-off score of 5 for the GDS has been well established and is typically used as the threshold score for defining depression with a sensitivity of 92% and a specificity of 81% [32].

Additional measures of interest included disease duration (calculated as study visit date minus the date of PD diagnosis in years), participant age, sex, race, and years of education.

Participant Selection and Characteristics

Participants in the current study were those in the PPMI PD cohort with a primary diagnosis of idiopathic PD. Given the primary aim of assessing the relationship between antidepressant use and ICD, participants missing medication log data (n = 61) and participants with a history of bipolar disorder (n = 4) were excluded. We excluded those with a history of bipolar disorder as antidepressants are known to have a role in provoking hypomania/mania, and the similarity between hypomania and ICD is challenging to separate clinically.

Statistical Analysis

As several variables were not normally distributed and some violated assumptions of equal variance, differences in participant characteristics and clinical outcomes were compared using two-sample Mann-Whitney tests for continuous variables and Pearson χ² tests for categorical variables. A Kaplan-Meier survival curve with a log-rank test was generated to assess the risk of ICD in those exposed to antidepressants prior to at the time of QUIP score measurement compared to those not exposed. Exposure was defined as receiving an antidepressant prior to and up until the visit date where the QUIP was measured. A QUIP greater than one constituted a positive ICD screen and was the outcome of interest in the survival analysis. Once a participant developed an ICD (QUIP>1) they were considered to have had the outcome of interest and exited the risk-set. A Cox proportional hazards regression model was constructed to calculate the hazard of ICD with antidepressant exposure adjusted for dopamine agonist exposure, MAOI-B exposure, amantadine exposure, depression, LEDD, disease duration, sex, and age. Adjustment variables were chosen based on previous literature establishing known ICD risk factors.[1, 5, 8, 33, 34] The proportional hazards assumption was inspected visually for each variable included in the final regression model, and a global test of the proportional hazard assumptions was conducted after assembling the final model to ensure that this assumption was not violated. Mixed effects logistic regression was also used to assess the odds of ICD within subjects with and without antidepressant exposure throughout the study period. Chi-square tests were also used to assess whether newly emergent ICD was associated with new prescriptions for serotonergic antidepressants. There were few missing cases; however, when missing items were encountered, they were excluded from statistical analyses. Alpha was set at 0.05, and a Holm-Bonferroni method for multiple comparisons was applied. STATA SE 18 (StataCorp LP, College Station, TX) was used for all analyses.

Results

Population Characteristics by Antidepressant Use

Table 1 displays study participants’ demographic, cognitive, and medication use patterns stratified by antidepressant use. A total of 1,045 individuals were included in the final analysis. The mean age was approximately 63 years. Sixty-one percent of participants were male, and 6% were non-white. Impulse control disorder was present in 10% of participants at visit one and in 27% of participants at some point during the study. Thirty-nine percent of participants were prescribed an antidepressant during the study. Participants were in the study for a mean of 7.2 years (median 7.9 years, range [1 –12.4 years]). The mean number of visits attended was 7.4 (median 8 visits, range [1 – 13]). Participants underwent a mean of 7.4 QUIP assessments (median 8, range [1– 13]).

Table 1:

Baseline Demographic, Cognitive, and Medication Use Information by Antidepressant Status

	All (n = 1,045)	Antidepressa nt − (n = 642)	Antidepressa nt + (n =403)	z-value/x2	Cohen’s D / Cramer’ s V	p-value
	Mean (SD)	Mean (SD)	Mean (SD)
Demographics
Age (years)	63.1 (9.6)	63.3 (9.5)	62.6 (9.7)	0.91	0.08	0.36
Sex (% Male)	60.6	64.3	54.6	9.8	0.10	0.002
Race (% non-white)	7.1	7.0	7.2	0.01	0.004	0.91
Education (years)	15.9 (3.5)	16.1 (3.4)	15.5 (3.7)	1.9	0.16	0.06
Disease duration at study entry (years)	1.4 (1.8)	1.4 (1.8)	1.5 (1.8)	−1.1	0.08	0.26
Cognition/Psychiatric Status
Geriatric Depression Scale (continuous score)	2.6 (2.8)	2.1 (2.5)	3.4 (3.0)	−8.4	0.49	<0.001
Geriatric Depression Scale (% ≥ 5)	18.0	12.3	27.1	36.5	0.19	<0.001
ICD (% yes during follow-up)	27.2	19.8	39.0	46.0	0.21	<0.001
Medication Use
Agonist Use (% Yes at baseline)	11.9	9.2	16.1	11.4	0.10	<0.001
Levodopa Equivalent Daily Dosage (mg/day at baseline)	145.0 (318.9)	125.0 (313.7)	176.9 (324.8)	−3.7	0.16	<0.001

^*p-values that maintained significance after the Holm-Bonferroni correction are bolded

^*Mann-Whitney tests were used for continuous variables and Pearson χ² tests were used for categorical variables. Degrees of freedom for all Pearson χ² tests were equal to one.

Participants on antidepressants compared to those not on antidepressants were more likely to be female (45% versus 36%, chi-square 9.8, df =1, p = 0.002) but similar with respect to age, race, years of education, and disease duration. Those on antidepressants had higher mean depression scores on the GDS at baseline (3.4 versus 2.1, z-value −8.4, p <0.001), were more likely to meet criteria for depression at baseline (27% versus 12 %, chi-square 36.5, df =1, p<0.001), and were more likely to have an ICD at some point during the study (39% versus 20%, chi-square 46.0, df =1, p<0.001) compared to those not on antidepressants. Participants on antidepressants were also more likely to be on a dopamine agonist at baseline (16.1% versus 9.2%, chi-square 11.4, df = 1, p = 0.001) and to have a higher LEDD at baseline (177mg/day versus 125mg/day, z-value −3.7, p<0.001) compared to those not on antidepressants.

Supplementary Table 1 displays the distribution of total QUIP scores across each visit as well as the distribution of ICD domains constituting positive screens on the QUIP at each visit.

Across the 52 sites, there were modest effect size differences in the rates of ICD (chi-square 110.9, df 51, Cramer’s V 0.15, p<0.001) and the rates of antidepressant exposure (chi-square 259.1, df 51, Cramer’s V 0.23, p<0.001). However, inclusion of site in regression models did not result in any meaningful change in model interpretation, and so participant site was ultimately not included in the final models.

Antidepressant Use and Risk of ICD

Figure 1 displays a Kaplan-Meier survival curve assessing the probability of ICD over time in those exposed to serotonergic antidepressants at any time prior to and up until QUIP score measurement compared to those not exposed to antidepressants prior to and at the time of QUIP score measurement. There was a significant increase in the probability of ICD in those exposed to antidepressants compared to those not exposed based on an equality of survivor functions log-rank chi-square test (chi-square 12.0, df =1, p<0.001). All classes of antidepressants (SSRIs, SNRIs, and other (mirtazapine, trazodone, vortioxetine, vilazodone, buspirone, atomoxetine) were associated with an increased risk of ICD based on chi-square tests with the exception of bupropion and tricyclic antidepressants (TCA).

Figure 1:

Kaplan Meier Survival Curve of ICD Over Time by Serotonergic Antidepressant Exposure Status

We also performed a chi-square test to assess whether newly emergent ICD was associated with new prescriptions for serotonergic antidepressants that were initiated after study entry but before the first emergence of an ICD. We found that those participants initiating new serotonergic antidepressants were more likely to have a newly emergent ICD (chi-square 370.1, df = 1, p<0.001). In addition, we assessed the relationship between length of antidepressant exposure and the occurrence of ICD. We found that new ICD was associated with a modestly longer period of antidepressant exposure (1.8 years versus 1.3 years, z-value −5.5, p <0.001, Cohen’s d 0.15), suggesting an increased risk with chronic exposure.

Cox Proportional Hazard Regression

Table 2 and Figure 2 display results from a Cox proportional hazard regression model assessing the hazard of ICD in those exposed to serotonergic antidepressants, adjusted for dopamine agonist use, amantadine, MAOI-B, bupropion, depression, LEDD, disease duration, sex, and age. Antidepressant use was associated with a hazard ratio for ICD of 1.4 (95% CI 1.0 – 1.8, z-value 2.1, p = 0.04). Dopamine agonist use, depression, and male sex were also significantly associated with an increased hazard of ICD. A model including an interaction term between antidepressant use and dopamine agonist use was tested; however, the interaction term coefficient was not significant, the other covariate coefficients did not change materially, and the model with an interaction term had a higher Akaike Information Criterion (AIC), and so the model without an interaction term was chosen.

Table 2:

Cox Proportional Hazard Regression Assessing Hazard of ICD with Antidepressant Use

Covariates	Hazard Ratio (CI)	z – value	p – value
Serotonergic Antidepressant (1 = Yes)	1.4 (1.0 – 1.8)	2.1	0.04
Dopamine Agonist (1 = Yes)	1.8 (1.3 – 2.4)	3.9	<0.001
GDS Score (>4 = 1)	2.1 (1.6 – 2.8)	5.1	<0.001
Bupropion (1 = Yes)	1.2 (0.6 – 2.1)	0.5	0.63
Amantadine (1 = Yes)	0.96 (0.54 – 1.7)	−0.1	0.90
MAOI-B (1 = Yes)	1.1 (0.83 – 1.5)	0.71	0.48
LEDD (mg/day)	1.0 (0.99 – 1.0)	1.5	0.14
Disease Duration (years)	1.0 (0.98 – 1.1)	1.3	0.18
Sex (1 = male, 0 = female)	1.5 (1.1 – 2.0)	2.7	0.006
Age (years)	0.99 (0.97 – 1.0)	−1.8	0.07

^*Displays the results of a multivariable cox proportional hazard regression model. Hazard ratios and test statistics are adjusted for all variables listed in the table

Figure 2:

Cox Proportional Hazard Regression Results - Antidepressant Use and ICD in PD

Mixed Effects Logistic Regression

Table 3 and Figure 3 display results of a mixed effects logistic regression model assessing the odds of ICD in those exposed to serotonergic antidepressants grouped by individual and adjusted for bupropion use, dopamine agonist use, amantadine use, MAOI use, depression, LEDD, disease duration, sex and age. Serotonergic antidepressant use was associated with an odds ratio for ICD of 1.9 (95% C.I. 1.4 – 2.7, z-value 3.9, p<0.001). Dopamine agonist use, MAOI use, depression, LEDD, disease duration, sex, and age were also associated with an increased odds of ICD.

Table 3:

Mixed Effects Logistic Regression Assessing Odds of ICD with Antidepressant Use

Covariates	Hazard Ratio (CI)	z – value	p – value
Serotonergic Antidepressant (1 = Yes)	1.9 (1.4 – 2.7)	3.9	<0.001
Dopamine Agonist (1 = Yes)	2.1 (1.6 – 2.8)	4.8	<0.001
GDS Score (>4 = 1)	2.4 (1.8 – 3.3)	5.8	<0.001
Bupropion (1 = Yes)	0.9 (0.4 – 2.0)	−0.2	0.81
Amantadine (1 = Yes)	0.9 (0.5 – 1.5)	−0.4	0.66
MAOI-B (1 = Yes)	1.7 (1.2 – 2.3)	3.2	0.002
LEDD (mg/day)	1.0 (1.0 – 1.0)	2.2	0.03
Disease Duration (years)	1.2 (1.2 – 1.3)	9.0	<0.001
Sex (1 = male, 0 = female)	1.9 (1.3 – 2.8)	3.1	0.002
Age (years)	1.0 (1.0 – 1.0)	−2.2	0.03

^*Displays the results of a multivariable mixed effects logistic regression model. Odds ratios and test statistics are adjusted for all variables listed in the table

Figure 3:

Mixed Effects Logistic Regression Results - Antidepressant Use and ICD in PD

A version of this data was previously presented orally and by poster at the Parkinson Study Group Annual Meeting in December 2024.

Discussion

Antidepressant use is associated with ICD in PD

This study found that serotonergic antidepressant use was associated with a significantly increased probability of ICD in PD after adjustment for several known ICD risk factors. Our results suggest that the increased risk of ICD associated with serotonergic antidepressant use is clinically significant, with nearly a forty-percent increased hazard of ICD in those exposed to serotonergic antidepressants. Consistent with prior studies, we also found that dopamine agonist use, depression, disease duration, and male sex were associated with an increased probability of ICD in PD. Importantly, our analysis suggests that bupropion, a commonly used non-serotonergic antidepressant, was not associated with an increased risk of ICD. However, there were relatively few participants taking bupropion prior to ICD assessments compared to those taking serotonergic antidepressants (4% versus 21%) and so it is possible that our analysis is underpowered to detect the true relationship between ICD and bupropion. Additionally, we found that participants with PD prescribed antidepressants are more likely to be female, have higher depression scores, are more likely to have an ICD, use dopamine agonists, and have a higher total LEDD than those not prescribed antidepressants.

Significance of Findings

Our findings are significant in their potential to impact the management of several clinical scenarios in PD. Being able to stratify patients based on their risk of developing ICD before treatment with a dopamine agonist could provide an opportunity to better tailor care and prevent challenging neuropsychiatric complications. Knowing that male patients on an antidepressant may be at a higher risk of ICD, for instance, could influence prescribing decisions about initiating dopamine agonists. Likewise, clinicians making treatment decisions for patients with depression/anxiety and co-existing ICD may consider alternatives to serotonergic antidepressants like bupropion, brain stimulation techniques, psychotherapy, or other novel treatment strategies. Additional studies are needed to prospectively determine the impact of bupropion on ICD, however, there is mixed evidence that bupropion may be associated with less mood cycling in bipolar disorder, potentially supporting its use in cases where ICD risk is high[35, 36]. In addition, bupropion has indications for compulsive-like behaviors including smoking cessation and weight loss, suggesting that it may be a safe alternative to serotonergic antidepressants[37, 38]. This ability to risk stratify patients proactively could aid in preventing some cases of ICD and may also support the development of new neurobiologically targeted therapy for depression and ICD in PD.

Antidepressants and ICD in PD

The impact of antidepressants on ICD risk in PD has not been a significant focus of prior studies; however, our study is not the first to suggest a link. In a large cross-sectional US cohort study of patients with PD, Jeon et al. found that those prescribed antidepressants in conjunction with a dopamine agonist had an adjusted odds ratio for ICD of 3.7 compared to those not receiving antidepressants [22]. Interestingly, in this analysis, the authors hypothesized that antidepressants would be protective against ICD and expressed surprise at finding an increased risk. Another study by Carbunaru et al. found that antidepressants significantly predicted higher scores on the Questionnaire for Impulsive-Compulsive Disorders in Parkinson’s Disease (QUIP) [23]. Our study expands upon these preliminary findings and strengthens the possibility of a causal connection between antidepressant use and ICD, given the longitudinal, time-to-event methodology utilized.

Neurobiology of ICD and Hypomania – Dopamine Dysregulation Hypothesis

Various studies have linked antidepressant treatment to changes in dopamine transmission, showing that chronic antidepressant use stimulates dopamine receptors and subsequent behavioral stimulant responses, leading to increased reward-related behaviors [39–41]. This could provide a neurobiological explanation for the link between ICD and antidepressant use.

Dopaminergic dysfunction, especially within the mesocorticolimbic pathways, has been linked to the emergence of ICD in PD. Some hypothesize that overstimulation of an intact ventral striatum contributes to ICD, especially in patients with early PD [5]. According to this theory, efforts to treat motor symptoms resulting from hypodopaminergic states in the dorsal striatum lead to excessive dopamine stimulation in the mesocorticolimbic pathways with subsequent dysregulation of reward pathways. A similar dopamine hypothesis underlies the origin of mood cycling in bipolar disorder [42]. Here, dysfunctions in dopaminergic homeostatic mechanisms are thought to give rise to periods of hyperdopaminergia and manic states, which are counterbalanced by periods of hypodopaminergia and depressive states. Rapid changes in dopamine tone are thought to lead to mood cycling in bipolar disorder, and a similar neurobiology may underly ICD in PD. The mechanism by which antidepressants provoke hypomanic/manic episodes in bipolar disorder remains to be clarified; however, changes in dopamine homeostasis are likely at play, which would explain the similar phenomena observed with ICD in PD.

There is also evidence that serotonin plays a key role in mediating reward-seeking behavior and that serotonin binding in the prefrontal cortex correlates positively with impulsivity and gambling disorder[43, 44]. Given this, serotonergic antidepressants may have a role in provoking ICD both through their impact on dopamine signaling as well as through direct serotonin modulation.

Strengths and Limitations

Our study is one of the first to perform a longitudinal time-to-event analysis assessing the association between antidepressant use and ICD in PD. While our study benefits from a large, geographically diverse, longitudinal dataset, some limitations merit consideration. The use of the QUIP in defining ICD is one limitation, as a definitive diagnosis of ICD requires a more detailed clinical assessment than the QUIP provides. The high sensitivity and specificity of the QUIP and its widespread use in the study of ICD help justify its use in our study; however, future prospective studies will benefit from the use of validated clinician assessments to confirm ICD diagnosis following a positive QUIP screen.

There were also modest site differences in rates of ICD and antidepressant use, however, there was no evidence that this meaningfully influenced the results of our final regression models. Future prospective analyses will benefit from exploration of site differences more comprehensively.

Another potential limitation is the lack of a measure of hypomanic/manic symptoms in our population. It is possible that some participants experienced hypomanic symptoms that overlap with ICD symptoms. However, the underlying neural correlates and behavioral phenomena appear to be similar in ICD and hypomania, and so it is possible that a significant group with ICD may also meet criteria for hypomania. Future prospective studies are needed to assess both symptom domains simultaneously to better ascertain the overlap in patients who become clinically hypomanic and those that experience ICD following exposure to antidepressants.

Finally, our study did not include structural or functional neuroimaging correlates of ICD in PD. Future studies mapping structural and functional correlates of these symptoms could support a dopamine dysregulation hypothesis and guide the development of future neurobiologically targeted therapies.

Conclusion

This study shows that antidepressant use is temporally associated with the emergence of ICD in patients with PD, even when adjusting for age, disease duration, and dopamine agonist use. This observation supports our hypothesis that the neurobiology giving rise to antidepressant-provoked hypomania/mania in bipolar disorder may be similar to the neurobiology underlying ICD in PD. Prospective studies are ultimately needed to expand upon our findings, confirm a causal link between antidepressant use and ICD, and clarify the neurobiological changes giving rise to these symptoms. While Parkinson’s disease is a challenging and progressive neurodegenerative illness, our study enhances the understanding of an important neuropsychiatric syndrome and promotes personalized treatment strategies to decrease the risk of ICD for individuals living with PD.

Highlights.

• What is the primary question addressed by this study?

  • This study explores the temporal association between antidepressant exposure and subsequent development of impulse control disorder (ICD) in patients with Parkinson’s disease (PD).
• What is the main finding of this study?

  • Antidepressant exposure is associated with a significantly increased risk of developing ICD in PD after adjusting for other established ICD risk factors including dopamine agonist use, depression, levodopa daily equivalent dosage (LEDD), disease duration, sex, and age.
• What is the meaning of the finding?

  • Our findings suggest that antidepressants may play a role in the development of ICD in some individuals with PD, and that a personalized approach to PD management is necessary to minimize the risk of this challenging neuropsychiatric condition.