Abstract
Background:
Response fluctuations and dyskinesias develop during the use of both levodopa (LD) and dopamine agonists (DA), but may not be equally disabling.
Objective:
To compare the risk and time of onset of disabling response fluctuations and dyskinesias (DRFD) among patients with Parkinson’s disease (PD) who were initially treated with either LD or DA.
Methods:
Open cohort study of all consecutive de-novo PD patients in routine clinical practice, included over a period of 15 years (median follow-up: 8.1 years, range 1.1–17.7), since embarking on LD or DA. Older patients and patients with more severe PD were started on LD (n = 77), younger patients on a DA (n = 50). Therapy was adjusted according to generally accepted guidelines. The primary endpoints were: the onset of response fluctuations, dyskinesias, and the moment when these complications became disabling (DRFD).
Results:
LD-starters developed response fluctuations 0.8 years earlier than DA-starters (p = 0.07), while dyskinesias appeared around 2.5 years earlier (p = 0.003). However, the risk and time of onset of DRFD did not differ statistically between the groups (LD-starters: 60% , median interval 7.3 years, DA-starters: 52% , 6.1 years, p = 0.63). DA-starters displayed a 0.19 points lower adjusted mean improvement in motor scores than LD-starters (p = 0.002). Adjustments for age and severity of PD at start of dopaminergic therapy did not change these results.
Conclusions:
In routine clinical practice, the risk and time of onset of DRFD is comparable for LD-starters versus DA-starters, but motor functioning is worse in DA-starters. These results support the use of LD as initial therapy for PD.
INTRODUCTION
In Parkinson’s disease (PD), initial treatment with dopamine agonists (DA) results in a lower incidence of response fluctuations and dyskinesias than treatment with levodopa plus a decarboxylase inhibitor (LD) [1]. With DA, however, the degree of symptom relief is lower while the incidence of short-term adverse motor events is higher [1]. Hence, most patients on DA monotherapy will eventually require LD rescue. Initially, response fluctuations and dyskinesias are not disabling and are straightforward to manage [2]. The moment when they become disabling has, therefore, more of an impact than the moment when they first appear. We report a prospective open-label cohort study comparing the risks of disabling response fluctuations and dyskinesias (DRFD) between patients who embarked on therapy with LD versus DA in routine clinical practice.
METHODS
Patients and drug regimes
Between 1989 and 2004, all consecutive de novo dopa-naïve PD patients from one neurologist’s practice (MH, movement disorders specialist) were treated according to generally accepted strategies [2, 3] (see ‘Background information on drug regimes’). In the current analysis, we excluded patients with a follow-up of <1 year, onset <40 years of age, or revision of diagnosis during follow-up. The diagnosis of PD met the UKPD Brain Bank criteria in all patients analyzed [4].
Patients were fully informed about therapy options for PD. Dopaminergic therapy was recommended once PD became disabling. Patients aged ≤65 years without clinically overt cognitive dysfunction and without manufacturers’ contraindications started with a DA unless rapid and optimal symptom control was needed. Patients aged ≥70 years started with LD. For patients between 65 and 70 years of age, the strategy was individually determined. The dose of each drug was gradually increased until a satisfactory response was obtained.
During the first years of this study, pergolide was the DA of first choice, bromocriptine was second choice. Ropinirole was first prescribed in 1998, pramipexole in 1999. When pergolide-associated heart disease was recognized, this drug was no longer prescribed to a new patient and only continued with appropriate cardiac monitoring. When necessary, LD-DA combination therapy was initiated in both treatment groups.
Patients were asked to report adverse events in a standard manner. Table 2 displays the adverse events restricted to those influencing anti-Parkinson therapy. Use of domperidone was permitted as a means of controlling dizziness, nausea or vomiting.
Dyskinesias and response fluctuations were detected by questioning and observation performed by MH. In case of any doubt, dyskinesias were observed on video or in the hospital. Patients were asked whether fluctuations or dyskinesias were disabling, i.e interfering with their independence or their planning of activities during the day. The moment when these fluctuations or dyskinesias could no longer be treated satisfactorily with adjustments of oral drug therapy (see ‘Background information on drug regimes’), they were labeled ‘disabling’
The severity of PD was assessed every 3 to 6 months using the Webster scale, the predecessor of the Unified Parkinson’s Disease Rating Scale (UPDRS [5]). The Webster scale, which was used because the study was started before the UPDRS became the standard has shown a high correlation (>0.8) with the UPDRS-motor part in all Hoehn and Yahr stages [6]. A state of disabling response fluctuations and dyskinesias probably equals a score of 2 (‘moderately disabling’) to 3 (‘severely disabling’) on UPDRS item 33.
Background information on drug regimes
In patients with minor disability, initial therapy was selegiline (Eldepryl, Viatris, Diemen, Netherlands) and amantadine (Symmetrel, Novartis, Arnhem, Netherlands), which had already frequently been prescribed by the referring physician. Selegiline was continued unless adverse events appeared. Amantadine was continued only if still proven effective after yearly withdrawals. The LD-group started with LD 100/25 mg 3td. When improvement became insufficient without being accompanied by late motor problems, the dose was gradually increased to a maximum of 2 tablets LD 100/25 mg 3td. The DA-group started with gradually increasing doses up to the patient’s satisfaction or maximum daily dose according to manufacturers’ guidelines. In case of (short term) adverse events that could not be managed with domperidone, the LD or DA dose was lowered and fixed on the maximum tolerated dose without adverse events. Patients with disabling adverse effects which could not be treated with domperidone or by lowering the DA dose, were advized to switch once to another DA (overnight, with the possibility of frequent consultations about adjustments of dose), or switch to LD, when preferred by the patient.
After the appearance of response fluctuations, the frequency of daily doses was increased to 4td in both groups. In doing so, the DA dose was increased beyond manufacturers’ maximal doses. Next, in those patients without selegiline, selegiline 5 mg 2td was tried first. When this failed in patients on LD, both as monotherapy and combination therapy, LD was changed into LD controlled-release (LD-CR). From 1999, LD-CR was replaced by the combination therapy entacapone plus LD. When LD-CR or entacapone were insufficient, they were discontinued. Therapy was continued by increasing frequencies of LD doses with adjustments of individual doses. In patients with monotherapy LD, DA were added when LD-CR, selegiline or entacapone failed.
Dyskinesias were treated by adjusting the doses and frequencies of LD, when necessary combined with lowering of selegiline, entacapone or the DA dose. In 1999, amantadine was introduced as an additional treatment option for dyskinesias, at a maximal dose of 100 mg 4td.
Statistical analysis
Demographic and drug therapy data were compared using a 2-tailed t-test and a chi-square test. All outcome measures were analyzed on an intention-to-treat basis. The risk of a long-term adverse event occurring was analyzed using the Cox proportional hazards model. Long-term adverse events were: response fluctuations, dyskinesias and disabling response fluctuations and/or dyskinesias (DRFD). A Kaplan-Meier curve was drawn to show the length of time until such an event occurred. When a patient’s follow-up ended without a particular event having occurred, the data were censored.
The adjusted mean change in Webster score from pre-dopaminergic treatment to last observation was analyzed using Mixed Model repeated measures with post-treatment Webster score as dependent variable, patient as random factor, LD-start versus DA-start as fixed factor, and pre-treatment Webster score and duration of therapy as covariates. The Webster score was analyzed until 20 patients remained in the follow-up (8 years) and until a duration of 5 years on DA monotherapy (9 patients remaining). To depict the change in Webster score, motor score averages were calculated over 6-month intervals.
Doses of LD-Controlled Release were multiplied by 0.75 [7] and analyzed together with LD. In case of entacapone use, we did not calculate adjustments in LD doses because entacapone allows only a relatively small reduction in LD dose and the frequency of entacapone use did not differ between the two groups.
RESULTS
Initial dopaminergic therapy
Table 1 presents demographic data related to drug therapy. One hundred twenty-seven patients were included in the analysis (77 on LD; 50 on DA). Six patients with a revised diagnosis were excluded. Two patients could not be treated with the preferred DA therapy because of recurrent psychosis and diabetic polyneuropathic hypotension and, therefore, started on LD. Eleven analyzed patients (8.6% ) were lost from follow-up before the study ended and before reaching the endpoint DRFD: one patient on pergolide died while driving a car, and ten patients preferred returning to their local hospital. Fig. 1A depicts the percentages of patients remaining in the follow-up over time.
LD-starters were older, with more severe PD, compared to DA-starters. However, there were no significant differences between the two groups regarding the use of non-dopaminergic anti-Parkinson medication. Eight DA-starters were still on monotherapy at the end of follow-up, with a median duration of 3.6 years (95% CI: 1.7–5.5; maximal: 9.1). The median duration of monotherapy of all DA-starters was 2.5 years (95% CI: 1.9–3.1). The median time to LD rescue was 2.3 years (95% CI: 1.6–3.0; maximum: 9.2). Twelve patients switched once to another DA, all without sufficient benefit. Eight patients discontinued DA and improved with LD. Thirty-four patients needed LD rescue because of adverse events or insufficient clinical improvement and improved thereafter. Among the LD-starters, 8 out of 28 patients with adjunct DA discontinued the latter. Table 2 summarizes the main reasons for changes in dopaminergic therapy. Short-term adverse effects of LD were responsive to domperidone and did not interfere with the dose of LD.
Response fluctuations and dyskinesias
Among DA-starters, the adjusted mean improvement in Webster score was 0.19 points lower than in LD-starters (95% CI: 0.32–0.06; p = 0.006). As long as the DA-starters remained on monotherapy, the difference was even 0.21 points (95% CI: 0.33–0.08; p < 0.001, Fig. 1B)
The time until the appearance of long-term adverse motor events is shown in Fig. 2. Response fluctuations occurred in 85.7% of the LD-starters, after a median of 3.1 years (95% CI: 2.5–3.7), and in 78% of the DA-starters, after a median of 3.9 years (95% CI: 3.5–4.3), with a hazard ratio (HR) of 1.43 (95% CI: 0.95–2.15, p = 0.078). Dyskinesias occurred in 74% of the LD-starters, after a median of 3.9 years (95% CI: 3.3–4.4), and in 50% of the DA-starters, after a median of 6.4 years (95% CI: 4.0–8.8), HR: 2.04 (95% CI: 1.26–3.30), p = 0.003. DRFD occurred in 59.7% of LD-starters, after a median of 7.3 years (95% CI: 6.1–8.4), and in 52% of DA-starters, after a median of 6.1 years (95% CI: 4.4–7.9); HR: 0.88 (95% CI: 0.54–1.44), p = 0.63. After adjustment for the confounders ‘age’ and ‘severity of PD’ at start of dopaminergic therapy, the statistical conclusions remained unchanged.
DISCUSSION
There is overwhelming evidence that initiating therapy with DA reduces the incidence of dyskinesias and response fluctuations [1]. This was also found in our cohort, drawn from routine clinical practice. The most important conclusion of our study is, however, that with adequate adjustments of therapy there is no difference in the length of time until fluctuations and dyskinesias become disabling, even in confounder-adjusted estimations. This appears in line with the findings in three large clinical trials. The 14-year PDRG-UK trial [8], which most closely resembles our cohort with patients broadly reflecting everyday practice, did not find a difference with respect to the risks of moderate and severe dyskinesias. Nor did the open long-term extensions of trials with ropinirole [9] and pramipexole [10]. These trials started with randomized groups, whereas in our cohort initial therapy was chosen individually according to clinical guidelines. Although the results from these heterogeneous studies can only be compared with caution, the results presented here seem to confirm that it does not matter whether dopaminergic therapy is initiated with LD or with DA as far as disabling response fluctuations and dyskinesias are concerned.
Similar to the majority of trials comparing DA with LD [1], we found greater improvements in motor performance in the patients who started on LD compared to DA (0.19 and 0.21 Webster points). In clinical trials, an improvement of 20% on the UPDRS motor part has been believed to be clinically meaningful [11]. In our cohort, the difference of 0.19 points in favor of LD equals an advantage of 22% of the baseline Webster score in the LD group (Table 1). So, according to this criterion, the difference in motor performance in favor of LD would be clinically meaningful.
Although the majority of patients (84% ) were treated with pergolide, various DA were used. There is, however, general consensus that all DA prescribed in this study have a similar clinical impact, and that motor improvement and the risks of response fluctuations and dyskinesias do not differ between the various DA [1].
In conclusion, the risks of disabling response fluctuations and dyskinesias do not differ between patients who started on LD or DA, treated in routine practice according to generally accepted guidelines. The improvement in motor performance with DA, however, is clearly less. Furthermore, when PD progresses into a state of significant non-dopaminergic pathology - which is not influenced by the initial treatment choice of DA or LD - the reduced frequency of fluctuations and dyskinesias in the DA group is not sustained [8, 12]. Overall, the results of our study in addition to those of others justify the use of LD as initial therapy in PD.
CONFLICTS OF INTEREST
The authors have no conflict of interest to report.
Footnotes
ACKNOWLEDGMENTS
We sincerely thank Michel Verbruggen (RUNMC, Nijmegen, The Netherlands) for technical support in the lay-out of the fig. C. Haaxma was supported by the Alkemade Fonds (research grant no. V6–02). B. Bloem received a VIDI grant from the Dutch Organization for Scientific Research (ZonMw, grant no. 016.076.352), another ZonMw research grant (no. 947.04.357) and a RUNMC research grant. M. Horstink was supported by research funding from Stichting De Regenboog. This study has not been sponsored by any farmaceutical company.