Subcutaneous Apomorphine Infusion Initiation Is Associated with Impulse Control Disorder Attenuation in Advanced Parkinson's Disease Patients: Insights from the French NS-Park Cohort

Abstract

Background: Impulse control disorders (ICD) are common non-motor complications in Parkinson's disease (PD), particularly in patients receiving oral dopamine agonists (DA). Continuous subcutaneous apomorphine infusion (CSAI) is a device-aided therapy for advanced PD, but its effects on ICD remain underexplored in real-world settings.

Objectives: To assess the impact of CSAI initiation on ICD prevalence and severity in a large real-world PD cohort and to compare ICD evolution in CSAI-treated patients versus orally-treated controls.

Methods: We analyzed data from the national prospective observational NS-Park cohort, selecting patients with documented ICD status before and after CSAI initiation. Changes in ICD prevalence and severity based on the MDS-UPDRS sub-item 1.6 were assessed using paired statistical tests, with additional sensitivity analyses based on time-restricted sub-cohorts (considering 60-, 24- and 12-months follow-up). A matched case-control analysis and a propensity score matching were used to compare CSAI-treated patients to orally-treated PD patients.

Results: 149 patients were included in the analysis. Before CSAI initiation, slight and mild/severe ICDs were present in 17% and 5% of the patients, respectively. After CSAI starting, ICD prevalence significantly decreased from 22% to 13%, (P = 0.003). These improvements were consistent across different time windows, despite an overall increase in DA levodopa-equivalent dose, with no associated mood worsening (up to 24-month follow-up). CSAI was associated with longitudinal ICD reduction, contrasting with the stable or worsening ICD trajectory in orally-treated controls, though trajectories were not statistically different.

Conclusions: The presented findings of our real-life cohort suggest that ICD tend to improve following CSAI initiation in patients with PD, likely due to a reduction of oral DA or the effect of continuous dopaminergic stimulation provided by the pump. While this observation is clinically relevant, it should be interpreted with caution given the study's observational design and the limitations inherent to using MDS-UPDRS sub-items for ICD assessment.

Keywords: Parkinson's disease; cohort; continuous‐subcutaneous apomorphine infusion; impulse control disorders.

Figure 1

Impulse control disorder (ICD) longitudinal changes. (A) Patient funnel with enrollment criteria for ICD assessment before and after continuous subcutaneous apomorphine infusion (CSAI) introduction. Each patient must have had at least one visit before and one visit after CSAI introduction with known ICD status. For all subsequent analyses in this figure, only one visit is taken per patient before and after CSAI introduction (the ones closest to the event) so that all the analyses are paired. Figures (B), (C), and D) have no temporal domain restrictions. (B) Exact number of patient with ICD = 0,1, or 2 on the evaluations closest to CSAI introduction (before and after). Exact scores distribution before vs after CSAI initiation were compare using ordered logistic regression. (C) Percentage of patient with ICD ≥1 before and after CSAI introduction. Paired/matched proportions compared using the McNemar's test. (D) Average ICD before and after CSAI introduction. Paired/matched scores compared using the Wilcoxon signed‐rank test. Figures (E), (F), (G), and (H) have both visits pre‐ and post‐CSAI initiation restricted to a 60‐month window from the event. (E) Median number of month from ICD evaluation and CSAI initiation or from CSAI initiation and subsequent ICD evaluation (60‐month window restricted). (F) Exact number of patient with ICD = 0,1, or 2 on the evaluations closest to CSAI introduction (before and after, 60‐month window restricted). Exact scores distribution before vs after CSAI initiation were compared using ordered logistic regression. (G) Percentage of patient with ICD ≥1 before and after CSAI introduction. Paired/matched proportions compared using the McNemar's test (60‐month window restricted). (H) Average ICD before and after CSAI introduction. Paired/matched scores compared using the Wilcoxon signed‐rank test (60‐month window restricted). Figures (I), (J), (K), and (L) have both visits pre‐ and post‐CSAI initiation restricted to a 24‐month window from the event. (I) Median number of month from ICD evaluation and CSAI initiation or from apomorphine initiation and subsequent ICD evaluation (24‐month window restricted). (J) Exact number of patient with ICD = 0,1, or 2 on the evaluations closest to CSAI introduction (before and after, 24‐month window restricted). Exact scores distribution before vs after CSAI initiation were compared using ordered logistic regression. (K) Percentage of patient with ICD ≥1 before and after CSAI introduction. Paired/matched proportions compared using the McNemar's test (24‐month window restricted). (L) Average ICD before and after CSAI introduction. Paired/matched scores compared using the Wilcoxon signed‐rank test (24‐month window restricted). Figures (M), (N), (O), and (P) have both visits pre‐ and post‐CSAI initiation restricted to a 12‐month window from the event. (M) Median number of month from ICD evaluation and CSAI initiation or from CSAI initiation and subsequent ICD evaluation (12‐month window restricted). (N) Exact number of patient with ICD = 0, 1, or 2 on the evaluations closest to CSAI introduction (before and after, 12‐month window restricted). Exact scores distribution before vs after CSAI initiation were compared using ordered logistic regression. (O) Percentage of patient with ICD ≥1 before and after CSAI introduction. Paired/matched proportions compared using the McNemar's test (12‐month window restricted). (P) Average ICD before and after CSAI introduction. Paired/matched scores compared using the Wilcoxon signed‐rank test (12‐month window restricted). ns non‐significant, *P‐value<0.05, **P‐value<0.01, ***P‐value<0.001.

Figure 2

Levodopa equivalent daily dose (LEDD) longitudinal changes. (A) Overall LEDD before vs after continuous subcutaneous apomorphine infusion (CSAI) introduction (entire cohort). (B) Overall LEDD before vs after CSAI introduction (patients with known non‐zero LEDD within the flaking 60 months). (C) Overall LEDD before vs after CSAI introduction (patients with known non‐zero LEDD within the flaking 24 months). (D) Overall LEDD before vs after CSAI introduction (patients with known non‐zero LEDD within the flaking 12 months). (E) Dopaminergic Agonist (DA) LEDD before vs after CSAI introduction (entire cohort). F) DA LEDD before vs after CSAI introduction (patients with known non‐zero LEDD within the flaking 60 months). (G) DA LEDD before vs after CSAI introduction (patients with known non‐zero LEDD within the flaking 24 months). (H) DA LEDD before vs after CSAI introduction (patients with known non‐zero LEDD within the flaking 12 months). ns non‐significant, *P‐value<0.05, **P‐value<0.01, ***P‐value<0.001.

Figure 3

Case control analysis. (A) Continuous subcutaneous apomorphine infusion (CSAI) levodopa equivalent daily dose (LEDD) and total dopamine agonists (DA) (including CSAI) LEDD distribution for all available patient‐visits with known/recorded doses (n = 677 patient‐visits post‐CSAI initiation). (B) Locally weighted regression fitting of Impulse control disorder (ICD) scores for patients who have been put on a CSAI (at some point during their trajectory) and respective control patients (1:5) over the sequential visit/evaluation number. No temporal restrictions applied. The control group was selected based on strict matching criteria to ensure comparability with the pump group. Specifically, a maximum of five control patients for each CSAI patient were selected. These control patients must have had the exact same gender, ON Hoehn & Yahr, and ICD score, age and disease duration ±5 years, on a date within 31 days of the CSAI introduction (for each case) that also corresponded to the same sequential patient visit at the center (±5 visits). While the trajectories seem to diverge, no statistically significant interactions was observed between the “group” and “time” variables upon linear effects modeling. (C) Love plot showing the balance of covariates before and after propensity score matching. The plot visualizes standardized mean differences for each covariate, where the matching procedure has improved the covariate balance between the treatment and control groups. (D) Bar plot of mean ICD scores with standard error comparing the treatment (CSAI) and control (never CSAI) groups. The bars represent the mean ICD score at each visit, with error bars indicating the standard error of the mean. To assess the treatment effect while accounting for potential confounding factors using propensity score matching, CSAI‐treated patients had their visits immediately before and after the initiation of treatment selected, ensuring that the treatment and control groups were well‐defined for comparison. For each of the treated patients, we matched five controls from the non‐treated group using a 1:5 nearest neighbor matching method. The propensity score was estimated using logistic regression, with treatment status as the dependent variable and a set of covariates as independent variables. These covariates included sex, age, disease duration, Hoehn & Yahr score, motor fluctuations, and dyskinesia, as well as the visit date. Notably, the exact visit date was also included in the propensity score calculation. The logistic regression model used a logit link function to calculate the propensity score, which represents the probability of receiving the treatment based on these covariates. Paired/matched scores compared using the Wilcoxon signed‐rank test (rank‐sum test in the case of independent samples—controls). ns non‐significant, *P‐value<0.05.