Introduction

Levodopa is the current gold standard therapy for the improvement of motor symptoms in patients with Parkinson’s disease (PD) [1]. However, long-term oral levodopa administration is associated with irregular, pulsatile dopamine stimulation and insufficient symptom control in patients with PD [2,3,4,5,6,7]. Progressive loss of striatal dopamine neurons and compromised dopamine signaling over time can contribute to greater motor complications in some patients with PD [2, 4]. Specifically, long-term use of oral levodopa (4–6 years) can be associated with up to a 40–75% likelihood of developing disabling motor complications and a 40% dyskinesia risk [1, 8, 9]. Further, the short levodopa half-life and irregular gastric emptying can lead to erratic plasma levodopa levels [2, 4, 10, 11]. Without consistent levodopa brain influx, patients are less likely to achieve a clinical response with oral levodopa over time [1]. Continuous dopaminergic stimulation produces several advantages over pulsatile dopaminergic stimulation [12]. Taken together, oral levodopa may not provide sufficient symptom control or clinical response as motor symptoms worsen over time in patients with PD.

Motor disturbances with long-term levodopa are also associated with a variety of nonmotor symptoms (NMS), all of which contribute to reduced health-related quality of life [1, 2, 13]. Treatment regimens include multiple medications aimed to increase dopamine neurotransmission (eg, addition of dopamine agonists) or to prolong the oral levodopa effect (eg, addition of monoamine oxidase [MAO] or catecholamine-O-methyltransferase [COMT] inhibitors) [14, 15]. These treatment regimens may provide symptom control for a period of time, but are often associated with drug-drug interactions, risk for medication errors, and reduced treatment adherence [14, 16, 17]. Additionally, add-on medications can have undesired side effects, such as impulse control disorders with dopamine agonists or diarrhea with COMT inhibitors [18, 19]. Likewise, medication non-adherence is prevalent in patients with PD, especially those with cognitive impairment or swallowing difficulties, leading to further motor dysfunction, as well as increased healthcare costs [1, 16, 20,21,22,23,24].

Levodopa–carbidopa intestinal gel (LCIG) allows individualized doses of levodopa to be infused continuously into the absorption site at the small intestine to maintain stable physiological dopamine levels [2, 25]. In clinical trials and real-world studies, LCIG leads to significantly greater improvements in PD symptom control and health-related quality of life, as compared to levodopa [2, 4, 26,27,28,29,30,31]. Likewise, a randomized, crossover design comparator study found that patients generally preferred infusion therapy with LCIG versus oral administration with levodopa (n = 24) [32]. Indeed, meta-analyses have found significant improvements in off-time and quality of life or comparable benefits with LCIG versus deep brain stimulation, continuous subcutaneous apomorphine infusion, and best medical treatment [33,34,35], and similar cost between LCIG and standard of care [36].

Studies on efficacy and safety supporting the use of LCIG vary in design, duration (3 weeks to 12 months), and patient population, with a few studies spanning over the course of years (2–5 years) [2, 4, 26,27,28,29,30]. More specifically, few studies have investigated the long-term use of LCIG monotherapy, with the longest durations being 2 years in the Global Long-term Registry on Efficacy and Safety of LCIG in patients with APD (GLORIA) study [37, 38], 3 years in the DUOGLOBE multinational real-world observational study [39], and in one study for up to 10 years (n = 37) [40]. There is a need for further clinical evidence on the long-term use of LCIG in patients with APD.

The COmedication Study assessing Mono- and cOmbination therapy with levodopa–carbidopa inteStinal gel (COSMOS) is a large, multinational study dedicated to evaluating the use of LCIG as a monotherapy or combination therapy [41]. Here, we report on long-term LCIG effects on motor symptoms, NMS, patient-reported outcomes, LCIG treatment settings, and safety according to LCIG treatment duration.

Methods

Study design

COSMOS was a multinational, retrospective, and cross-sectional post-marketing observational study in patients with APD treated with LCIG in routine clinical care (Clinicaltrials.gov identifier: NCT03362879; Fasano et al. [41] for full methods). Briefly, data were collected retrospectively via chart review and cross-sectionally at a single study patient visit. Clinical data were entered into a web-based electronic data capture system for analysis.

Participants

Patients were eligible for inclusion if they had a diagnosis of APD and had received ongoing LCIG treatment for ≥ 12 months prior to the study visit. Patients were excluded if they had participated in a concurrent or prior clinical trial involving LCIG or were unable to complete the study questionnaire. Written informed consent was obtained by each patient or legally authorized representative prior to any data collection. For this analysis, patients were stratified into five groups based on LCIG treatment duration at the patient visit: 1–2 years LCIG (> 1 to ≤ 2 years LCIG), 2–3 years LCIG (> 2 to ≤ 3 years LCIG), 3–4 years LCIG (> 3 to ≤ 4 years LCIG), 4–5 years LCIG (> 4 to ≤ 5 years LCIG), and > 5 years LCIG.

Demographics and clinical characteristics

During the patient visit, physicians collected current patient demographic information and medical history. Medical history data included PD history, clinical PD status, and LCIG treatment settings.

Clinical assessments

At baseline, defined as immediately prior to LCIG initiation and at the patient visit, physicians used the Unified Parkinson’s Disease Rating Scale (UPDRS) to measure “off” time (Part IV item 39, modified) and dyskinesia duration (Part IV item 32, modified), as well as dyskinesia severity (Part IV item 33). The UPDRS was selected based on the feasibility of completion within the clinical practice setting. “Off” time and dyskinesia duration were documented as hours during the day prior to the clinical visit as reported by the patient. At the patient visit, physicians assessed the prevalence, severity rating (none, mild, moderate, or severe), and frequency rating (rarely, often, frequent, or very frequent) of motor symptoms, NMS, and symptoms related to treatment from both timepoints (at baseline and at the patient visit). Symptoms were defined as characterized by the PD composite scale [42]. In addition, patients were assessed with the NMS Scale (NMSS), PD Sleep Scale Version 2 (PDSS-2), and PD Quality of Life Questionnaire (PDQ-8) at the patient visit. LCIG dosing and infusion-related dosing parameters were collected from medical records and at the patient visit. Data on add-on medications were collected.

Safety

Data from safety assessments were previously collected by healthcare professionals and documented in medical records along with any prior suspected adverse reactions from the patients’ medical files. Adverse events (AEs) possibly related to treatment or device were documented.

Statistical analysis

Data from medical records and patient visits were analyzed using descriptive statistics. All statistical analyses were carried out using SAS® version 9.4 (SAS Institute, Inc., Cary, NC, USA). Prevalence, defined as the percentage of patients affected, was analyzed for motor symptoms, NMS, and symptoms related to treatment. For motor symptoms, NMS, and treatment-related symptoms, ratings of frequency and severity were transformed into numerical scores and analyzed. For each symptom, variables ‘severity’ and ‘frequency’ were transformed as follows: for the transformed severity variable 'no symptom' was 0, mild = 1, moderate = 2, and severe = 3, and unknown was set to missing. For the transformed frequency variable 'no symptom' was 0, rarely = 1, often = 2, frequent = 3, very frequent = 4, and unknown was set to missing. A chi-squared test was used to compare categorical data (eg, PD motor phenotype, monotherapy, and motor symptom and NMS prevalence when possible). The Kappa test was used to analyze motor symptoms and NMS prevalence (within-group differences). The Wilcoxon test was used to analyze quantitative comparisons of continuous, non-normally distributed data (eg, between-group and within-group differences of severity and frequency scores for NMSS, PDSS-2, and PDQ-8).

Results

Participants

The COSMOS study included 409 patients from 49 sites in 14 countries [41]. A total of 387 patients were included in this retrospective analysis based on data availability from medical records at the patient visit date (ie, 22 patients were omitted due to lack of data on the LCIG initiation date in case report forms and the inability to calculate LCIG duration). The number of patients per LCIG duration were as follows: 1–2 years LCIG (n = 156); 2–3 years LCIG (n = 80); 3–4 years LCIG (n = 61); 4–5 years LCIG (n = 30); > 5 years LCIG (n = 60). Demographics and clinical characteristics at baseline were generally similar among groups (Table 1). The majority of patients were male with a mean ± standard deviation (SD) age of patients at baseline ranging from 64.5 ± 7.6 years of age (4–5 years LCIG) to 67.9 ± 7.4 years of age (1–2 years LCIG) (Table 1). Across all groups, the most common reason for LCIG initiation was due to disabling motor fluctuations/off periods (≥ 83.3%) (Table 1). The other most common reasons for LCIG initiation were decreased quality of life, uncontrolled dyskinesia, and lack of efficacy of previous treatment. The time from PD diagnosis to LCIG initiation ranged from a mean ± SD of 12.1 ± 5.4 years (1–2 years LCIG) to 13.8 ± 5.1 years (> 5 years LCIG) (Table 1). At baseline, patients in the 4–5 years LCIG group had lower mean ± SD dyskinesia duration (2.7 ± 2.6 h) as compared to other groups, ranging from mean ± SD of 3.1 ± 2.9 to 4.9 ± 4.5 h (Table 1).

Table 1 Baseline demographics and clinical characteristics of patients according to duration of LCIG treatment
Full size table

Clinical assessments

“Off” time, “on” time, and dyskinesia severity

At the time of the patient visit, the duration of “off” time was reduced from baseline in all groups (p < 0.001) (Fig. 1). The duration of “on” time with dyskinesia was reduced from baseline in all (p < 0.001), with the exception of the 4–5 years LCIG group (p = 0.1378). “On” time without dyskinesia was increased from baseline in all groups (p < 0.0001 for all groups except 4–5 years LCIG [p = 0.0002]) (Fig. 1). Reductions from baseline were found in dyskinesia severity: mean ± SD, ranging from − 0.7 ± 1.3 to − 0.9 ± 1.3 (p < 0.0001 for all groups except 4–5 years LCIG [p = 0.0284] and > 5 years LCIG [p = 0.0003]).

Fig. 1
figure 1

Change from Baseline of “Off” Time and “On” Time with and without Dyskinesia During Waking Hours from LCIG Initiation to Patient Visit According to Duration of LCIG Treatment. Significance for change from baseline as follows: NS: p > 0.05; **p < 0.001; ***p < 0.0001. LCIG levodopa–carbidopa intestinal gel, NS not significant, SD standard deviation

Full size image

Motor symptoms, NMS, and treatment-related symptoms

For many of the individual motor symptoms, reductions from baseline in prevalence were found in all groups for tremor and nocturnal/morning akinesia, while dysphagia increased (all groups except 4–5 years LCIG) and hypophonia increased (4–5 years LCIG only) (Fig. 2A). Some individual NMS showed similar reductions from baseline in all groups for the prevalence of anxiety, pain, constipation, gambling, and dopamine dysregulation syndrome, with the exception of constipation (> 5 years LCIG) and dopamine dysregulation syndrome (1–2 years LCIG and > 5 years LCIG) (Fig. 2B). There were no significant changes from baseline in the prevalence of treatment-related symptoms (Supplemental Fig. 1).

Fig. 2
figure 2

Change from Baseline in Prevalence of Motor Symptoms (A) and NMS (B). According to Duration of LCIG Treatment. Negative values indicate a decrease in symptom prevalence. LCIG levodopa–carbidopa intestinal gel, NMS non-motor symptoms

Full size image

Reductions from baseline in severity and frequency were seen in individual motor symptoms, specifically for bradykinesia, rigidity, tremor, dystonia/cramps, gait impairment, nocturnal/morning akinesia, and freezing of gait for all groups (Supplemental Figs. 2A and 3A). Similarly, patients in all groups had reductions from baseline in the severity and frequency of some of the individual NMS and treatment-related symptoms assessed (Supplemental Figs. 2B, C and 3B, C). Differences between groups are noted in Supplemental Figs. 2 and 3.

NMSS, PDSS-2, and PDQ-8

NMSS scores were lower with 1–2 and 2–3 years LCIG vs > 5 years LCIG (mean ± SD, 1–2 years LCIG [55.3 ± 43.8]; 2–3 years LCIG [52.6 ± 38.3] vs > 5 years LCIG [68.9 ± 37.4], p = 0.0068 and p = 0.0057) (Table 2). PDSS-2 total scores were similar across groups at the patient visit (mean ± SD, ranging from 18.9 ± 9.1 to 22.0 ± 11.9) (Table 2). PDQ-8 was lower with 1–2 years and 2–3 years LCIG as compared to > 5 years LCIG (mean ± SD, 1–2 years LCIG [36.5 ± 19.0] and 2–3 years LCIG [38.9 ± 18.5] vs > 5 years LCIG [46.3 ± 16.5]; p = 0.0003 and p = 0.0076) (Table 2). Differences between groups with p < 0.05 are as follows: NMSS, 1–2 years vs > 5 years (p = 0.0068) and 2–3 years vs > 5 years (p = 0.0057); PDQ-8, 1–2 years vs 3–4 years (p = 0.0052), 1–2 years vs > 5 years (p = 0.0003), 2–3 years vs > 5 years (p = 0.0076). No differences between groups were found in PDSS-2.

Table 2 NMSS, PDSS, and PDQ-8 scores assessed via questionnaires according to duration of LCIG treatment
Full size table

LCIG dosage and add-on medications

LCIG total doses and LEDD were similar across groups at both LCIG initiation and at the patient visit (Table 3). Add-on medications were reduced in most groups at the patient visit as compared to baseline (Fig. 3A). Additionally, the number of add-on medication intakes were reduced at the patient visit as compared to baseline, independent of LCIG duration (p < 0.0001) (Fig. 3B). At 12 months after LCIG initiation, the percentages of patients receiving LCIG as monotherapy ranged from 22.4 to 46.7%, while the percentages of those receiving LCIG monotherapy plus night medication ranged from 10.0 to 30.3% and those receiving polytherapy ranged from 40.1 to 52.6% (Table 3).

Table 3 LCIG settings and LEDD at baseline and at the patient visit according to duration of LCIG treatment
Full size table
Fig. 3
figure 3

Percentage of Patients Receiving Add-on Medications (A) and Add-on PD Medication Intakes (B) at Baseline and Patient Visita. aLess than 10% of patients had any intake of apomorphine, anticholinergics, and other add-on medications (not shown). No changes were found with NMDA antagonists as add-on medications (not shown). bLevodopa treatment included nocturnal levodopa, among other reasons, at the patient visit. COMT catecholamine-O-methyltransferase, LCIG levodopa–carbidopa intestinal gel, MAO monoamine oxidase, NMDA N-methyl-d-aspartate, PD Parkinson’s disease

Full size image

Safety

A total of 109 AEs of any type were recorded via medical records during LCIG initiation and LCIG maintenance treatment for 387 patients, ranging across groups from 24% (37/156) in 1–2 years LCIG to 36% (22/61) in 3–4 years LCIG. The most common AEs reported were stoma site infection (> 5 years LCIG: 4/60 [7%] and 3–4 years LCIG: 3/61 [5%]) and unintentional medical device removal (3–4 years LCIG: 3/61 [5%]).

Discussion

In this post hoc analysis of the retrospective, cross-sectional, multinational COSMOS study, we reported LCIG treatment benefits present in the short-term and sustained through long-term LCIG treatment. Motor fluctuations, “off” time, dyskinesia, and motor symptoms improved from baseline for all groups, independent of LCIG duration. With continuous levodopa infusion and subsequent continuous dopaminergic stimulation, patients were able to maintain symptom control beyond 5 years with a stable levodopa requirement while reducing the burden of a complex add-on medication regimen. Approximately 20–45% of patients were on LCIG monotherapy at least 12 months after LCIG initiation, with 39.7% receiving LCIG monotherapy (and another 17.2% receiving LCIG monotherapy plus nighttime medication) after more than 5 years of treatment. The improvement of dyskinesia duration and severity, as well as “off” time, and the large percentage of patients on LCIG monotherapy, as found in the COSMOS primary analysis [41], support the potential use of LCIG for long-term symptom control.

A major strength of this study was the use of real-world, multinational, large-cohort clinical data. This post-hoc analysis of the COSMOS study is one of the few analyses to evaluate long-term LCIG therapy in patients with APD, and the first to systematically compare outcomes according to treatment duration, as well as long-term data for the rate of monotherapy, LEDD, and a number of intakes. The improvements in motor symptoms and NMS are supported by previous observational studies, as well as randomized controlled trials that have identified improvements in motor complications and NMS associated with LCIG treatment [2, 4, 26,27,28,29,30, 38].

The study included a partially retrospective design and observational analyses, without randomization at the study start, which limited data collection and interpretation. In addition, this study included only patients treated with ongoing LCIG who were able to sustain LCIG treatment for at least 12 months. Thus, the results of this analysis are not representative of all patients who initiate LCIG, possibly introducing an unintended bias. However, given the nature of the study, it is conceivable that patients with suboptimal outcomes over time discontinued LCIG and were not included in the analysis. In addition, recall bias at the patient visit may also influence the data upon which this post hoc analysis is based, wherein events that happened long ago are more likely to be misreported and thus potentially introducing systematic error across the treatment duration groups. Finally, the study was unable to achieve extended, long-term follow-up much beyond 5 years due to the low number of patients tracked for longer durations. However, this study design allows for a larger patient cohort than one would be able to track in a long-term treatment prospective design.

The data from this real-world analysis of patients with APD demonstrate patient outcome profiles with long-term LCIG treatment. Treatment with LCIG was associated with a sustained improvement in both dyskinesia duration and severity, independent of LCIG treatment duration. Stable levodopa delivery via LCIG reduced add-on medications. These results are in line with the EAN MDS guidelines for invasive therapies which identified improved time and quality of life in patients with LCIG versus oral therapy and shed light on potential patient selection for long-term LCIG treatment [43]. LCIG may be a long-term solution for carefully selected patients based on patient preferences and clinical characteristics (e.g., those without dementia) such that the benefits outweigh any potential risks [34, 44, 45]. Due to the long-term progressive nature of PD, as well as the aging process and potential treatment risks, patients with APD receiving treatments like LCIG may require consistent monitoring for worsening symptoms related to cognitive decline, vitamin B12 deficiency, or polyneuropathy [46, 47].

The study provides evidence that supports LCIG as an option to maintain adequate symptom control in the long term, despite the continued progression of PD. Adverse events were consistent with the known LCIG safety profile [2, 26, 41, 48]. Furthermore, LCIG can be a long-term solution for patients with memory or swallowing difficulties associated with APD who follow multi-faceted therapeutic regimens requiring many pills per day and may even eliminate the need to use add-on medications, as monotherapy was an option for many patients. Full understanding of the potential long-term benefits, beyond symptom control, with continuous dopaminergic stimulation via continuous levodopa delivery will require additional studies. In conclusion, long-term use of LCIG maintains reductions in “off” time, dyskinesia duration and severity, reduces the burden of several motor symptoms and NMS, and lessens the need for add-on medications, with a stable requirement of LCIG.