Long-term Effect of Initiating Pramipexole vs Levodopa in Early Parkinson Disease FREE

Arguments in favor of starting a dopamine agonist rather than levodopa in early Parkinson disease (PD) have been advocated on the belief that delaying dopaminergic events (primarily dyskinesias and wearing off) has benefits in terms of long-term disability and quality of life. Indeed, several phase 4 randomized controlled trials have been conducted and published using the time until development of dopaminergic events as the primary outcome.^{1 - 3} These studies all show that the strategy of initial treatment with dopamine agonists delays the onset of dyskinesias and wearing off, whereas initial treatment with levodopa is generally associated with greater improvement in disease symptoms as measured by the Unified Parkinson's Disease Rating Scale (UPDRS). As a result, clinical practice guidelines have concluded that both are options for initial management despite limited data on the long-term implications of either treatment approach.⁴

Long-term follow-up studies comparing initial treatment with dopamine agonists vs levodopa in early PD have been few.^{5 - 9} In an open-label, 10-year naturalistic follow-up study comparing individuals randomized to initial treatment with ropinirole hydrochloride vs initial treatment with levodopa, motor complications were less common in the initial ropinirole group and no significant differences were seen in measures of disease severity, disability, quality of life, or somnolence⁸ ; however, only 18% of the original cohort was evaluated. In another randomized open trial, after a median of 14 years of follow-up, subjects randomized to initial bromocriptine mesylate had greater disability and were no different in terms of the prevalence of dyskinesias, motor fluctuations, dementia, or mortality compared with subjects randomized to initial levodopa⁹ ; again, only a small fraction of the surviving cohort (21%) was evaluated.

Here we report the results of the CALM Cohort study, an extended follow-up of the Comparison of the Agonist Pramipexole With Levodopa on Motor Complications of Parkinson's Disease (CALM-PD) trial participants using outcomes from multiple domains, to better understand the long-term consequences of initial treatment with either pramipexole dihydrochloride or levodopa in individuals with early PD.

The CALM Cohort study extended the follow-up of subjects enrolled in the CALM-PD clinical trial. The detailed methods and results of the CALM-PD trial have been reported elsewhere.^{1 ,10 - 11} The CALM-PD trial enrolled subjects between October 1996 and August 1997 at 22 sites in the United States (17 sites) and Canada (5 sites); the subjects had idiopathic PD for less than 7 years and required dopaminergic antiparkinsonian therapy at the time of enrollment. Patients who had taken levodopa or a dopaminergic agonist in the 2 months prior to enrollment were excluded.

Eligible patients were randomized 1:1 to initial treatment with either pramipexole alone or levodopa in combination with carbidopa, and the treatment was titrated to 1 of 3 dosage levels over a period of 10 weeks depending on need (level 1: 1.5 mg/d of pramipexole dihydrochloride, or 75 mg/d of carbidopa and 300 mg/d of levodopa; level 2: 3.0 mg/d of pramipexole dihydrochloride, or 112.5 mg/d of carbidopa and 450 mg/d of levodopa; and level 3: 4.5 mg/d of pramipexole dihydrochloride, or 150 mg/d of carbidopa and 600 mg/d of levodopa). Subjects were evaluated every 3 months until the last enrolled subject completed a 48-month visit. During the follow-up period, subjects with emerging disability could be treated with open-label levodopa. In the final year of the trial, subjects who developed wearing off, dyskinesias, or on-off fluctuations were permitted to do the following: (1) increase or decrease study drug dosage levels; (2) add sustained-release carbidopa and levodopa or a catechol O-methyltransferase inhibitor; and (3) alter or add amantadine hydrochloride, anticholinergic medications, or selegiline hydrochloride. Treatment assignments were disclosed to subjects between June and August 2001, and the patients' subsequent care was left to the discretion of the treating neurologist.

After the CALM-PD trial ended, patients were offered entry into the CALM Cohort study and consenting subjects were enrolled from April 2002 to January 2004. Effort was made to contact all of the patients, including those who completed the 4-year trial as well as those who prematurely withdrew. Patients were followed up annually for 2 years, for a mean (SD) of 6.0 (0.2) years of follow-up since randomization. The study was reviewed and approved by the institutional review board at each of the participating sites and all of the subjects gave written consent to participate. The following outcomes were assessed at the 5- and 6-year annual face-to-face visits.

Disability was assessed using the modified Schwab and England Activities of Daily Living (S/E ADL) Scale.¹² This scale assesses a subject's capacity to perform ADL as a percentage of normal (0%-100%). The subject and spouse were instructed to indicate the score, to the nearest 1%, that described the subject's condition during the week prior to assessment. Direct and indirect methods of obtaining an S/E ADL Scale score were used. Using the direct method, subjects were asked to provide an overall self-reported estimate of the modified S/E ADL Scale disability score in the week prior to the study visit. Using the indirect method, we asked subjects to estimate the proportion of their waking day in the “on” state (ie, when medications are working relatively well). We then asked subjects to estimate their S/E ADL Scale score during on times and off times. Using both the disability and time estimates, a weighted average S/E ADL Scale score was calculated. The S/E ADL Scale score derived using the indirect method served as the primary outcome variable.

We assessed for the presence of dyskinesias, wearing off, on-off effects, and freezing as defined in a prior article.¹ The severity of these events was assessed using part IV of the UPDRS. The Lang-Fahn ADL dyskinesia scale was used to rate the effect of the subject's worst dyskinesias on the following 5 ADL: (1) handwriting and drawing; (2) cutting food and handling utensils; (3) dressing; (4) hygiene; and (5) walking.¹³

We assessed disease severity using parts I (mental), II (ADL), and III (motor) of the UPDRS. The UPDRS is a standardized, reliable, and valid instrument for assessing the severity of the clinical features of PD.¹⁴

We measured daytime sleepiness with the Epworth Sleepiness Scale (ESS).¹⁵ The ESS is a simple, 8-item, self-administered questionnaire asking subjects to rate on a scale of 0 to 3 how likely they would be to doze off or fall asleep in 8 situations. A higher score is associated with increased sleepiness, and a score of 10 or more is considered excessively sleepy in persons with sleep disorders¹⁶ and has been associated with “sleep attacks” in PD.¹⁷ The ESS has shown excellent internal consistency and reliability and can distinguish subjects with mild, moderate, and severe daytime sleepiness.¹⁸

We assessed the presence and effect of edema using the following 2 questions. First, does the subject have edema (peripheral, generalized, tongue, lymphedema)? If yes, does the edema interfere with function or ambulation?

We used the Geriatric Depression Rating Scale Short Form to measure depressive symptoms.¹⁹ The Geriatric Depression Rating Scale Short Form has been shown to be a useful measure of depressive symptoms in elderly and nonelderly groups, including subjects with PD.²⁰

We performed cognitive assessments with the Mini-Mental State Examination, an 11-item, well-validated tool for cognitive screening.²¹

One disease-specific and 2 generic quality-of-life instruments were used. The Parkinson's Disease Quality-of-Life Scale is a PD-specific quality-of-life scale that consists of 32 items and is scored on a 100-point scale. It includes 7 domains: social and role function, self-image and sexuality, sleep, outlook, physical function, independence, and urinary function. Factor structure, reliability, and validity of the scale have been established in a cross-sectional study.²² The EuroQol EQ-5D is a generic, preference-based, multidimensional measure of health-related quality of life that can be expressed as a single index value.^{23 - 24} It is defined over 5 domains of health (mobility, self-care, usual activity, pain, and emotional health). It also includes a visual analog (thermometer-type) rating scale on which the subject rates his or her current health state on a scale from 0 to 100. The EQ-5D has been shown to be a feasible and valid instrument to measure quality of life in subjects with PD and reflects the severity and complications of disease.²⁴ The 12-Item Short Form Health Survey is a widely used measure of generic health-related quality of life.²⁵ The instrument yields 2 summary scores characterizing the physical and mental health components. These range from 0 to 100, with higher scores indicating better quality of life.

All of the PD medications and dosages were collected at the final visit.

The primary statistical analyses included only subjects who enrolled in the CALM Cohort study (n = 222). All of the statistical tests were 2-tailed and were performed using a significance level of 5%. The analyses focused on comparisons between the initial pramipexole and initial levodopa groups at the last available subject visit. Because this was a subset of the originally randomized full cohort of subjects (N = 301), the treatment groups could not necessarily be considered to be comparable by randomization. We therefore used propensity score adjustment^{26 - 27} to minimize the potential bias due to noncomparability. The propensity score for a subject is the estimated probability of being assigned to receive initial pramipexole given the subject's baseline characteristics. This was determined for each subject as the predicted probability of assignment to initial pramipexole using a logistic regression model that included the following characteristics at randomization as independent variables: age, sex, race, years since PD diagnosis, history of levodopa exposure, current use of selegiline, current use of amantadine, current use of anticholinergic medication, employment status (full-time, part-time, unemployed), total UPDRS score, Hoehn and Yahr stage, Mini-Mental State Examination score, Parkinson's Disease Quality-of-Life Scale total score, comorbid cardiac illness, and presence of comorbid illness in more than 5 body systems (of 11 queried).

The analysis of the primary outcome variable used an analysis of covariance model that included treatment group as the factor of interest and enrolling investigator and propensity score as covariates. The propensity score was categorized using quintiles for purposes of these analyses.²⁶ A 95% confidence interval was computed for the difference between the adjusted group means. Analyses of continuous secondary outcome variables were similarly performed except that the change from baseline was used as the outcome variable if the outcome was measured at the original CALM-PD baseline; in this case, the analysis of covariance model also included the baseline value of the outcome variable as a covariate. Interactions between treatment and enrolling investigator were tested by including the appropriate interaction terms in the model, but no such interactions were found. Analyses of dichotomous outcome variables were similar except that a logistic regression model that included treatment group and propensity score as independent variables was used.

The Figureshows subject flow from the start of the CALM-PD trial to the completion of the CALM Cohort study. Of the 301 subjects initially randomized, 151 were assigned to the initial pramipexole arm and 150 were assigned to the levodopa arm. Sixty-one percent of subjects (183 of 301 enrolled) completed 4 years of follow-up in the CALM-PD trial. A total of 222 subjects were recruited into the CALM Cohort, including 158 of the 183 who completed the CALM-PD trial and 64 of the 118 who had withdrawn prematurely from the CALM-PD trial. Of the 79 subjects not recruited into the CALM Cohort, 25 died, 37 refused to participate, and 17 were lost to follow-up. The mean (SD) duration of follow-up after the original randomization was 6.0 (0.2) years (range, 4.5-6.8 years).

Table 1shows the baseline characteristics, at enrollment in the CALM-PD trial, of the 222 subjects (108 in the pramipexole group and 114 in the levodopa group) who entered the CALM Cohort study by initial treatment assignment. The 2 groups in the cohort were similar across most baseline characteristics except that there were lower quality-of-life scores in the pramipexole group, a difference seen in the original randomization.¹ The pramipexole group also tended to have more women and a higher initial study medication dosage level. Patients who did not enter the cohort were older, had more severe disease, and had worse quality of life.

Table 2shows the PD medications and mean dosages used by study subjects at the final visit, approximately 1 to 2 years after the breaking of the study blind. More than 90% of patients in each initial treatment group received concomitant levodopa therapy. More than 80% of those originally randomized to initial pramipexole were still taking an agonist, most commonly pramipexole, compared with fewer than 50% of the subjects in the initial levodopa group (P < .001). The lower frequency of agonist use in the initial levodopa group, however, was partially offset by a higher use of other PD medications (particularly selegiline and amantadine).

Table 3shows comparative outcome data at the final visit by original treatment assignment. Mean self-reported disability as measured by either the weighted average of S/E ADL Scale scores in the on and off states (indirect method) or the subject's overall S/E ADL Scale score (direct method) was similar in the 2 treatment groups. At the final visit, dyskinesias and wearing off were more common in the initial levodopa group than in the initial pramipexole group (dyskinesias: 36.8% vs 20.4%, respectively; P = .004; wearing off: 58.8% vs 44.4%, respectively; P = .01); freezing was slightly more common in the initial pramipexole group than in the initial levodopa group (34.7% vs 26.2%, respectively), but this difference was not statistically significant (P = .30). The initial pramipexole group as compared with the initial levodopa group tended to have a higher mean (SD) percentage of the waking day in the on state (85.6% [21.2%] vs 83.9% [17.8%], respectively; P = .24) and a lower mean (SD) score on the Lang-Fahn ADL dyskinesia scale rating the disability associated with dyskinesias (1.1 [2.9] vs 1.3 [3.2], respectively; P = .06), but these differences were not statistically significant. These differences may in part reflect the lower incidences of wearing off and dyskinesias in the group initially treated with pramipexole. Only 7 subjects (3 in the initial pramipexole group and 4 in the initial levodopa group) reported dyskinesias that were at least moderately disabling, and only 10 subjects (6 in the initial pramipexole group and 4 in the initial levodopa group) reported having painful dyskinesias at the final visit.

The mean changes from baseline in the UPDRS total, motor, and ADL scores favored the group initially treated with levodopa, but the magnitude of benefit was less than that seen in the 4-year clinical trial and was not statistically significant (Table 3). Regarding other outcomes, the mean (SD) ESS score at the final visit was significantly higher (indicating more sleepiness) in the initial pramipexole group compared with that in the initial levodopa group (11.3 [5.8] vs 8.6 [4.7], respectively; P < .001); likewise, the percentage of subjects with an ESS score of 10 or higher was 57.4% in the initial pramipexole group compared with 35.2% in the initial levodopa group (P = .002). The prevalence of edema at the final visit was higher in the initial pramipexole group (27.1%) than in the initial levodopa group (14.4%) (P = .04), but functionally significant edema was rare in both groups (Table 3). Cognitive, mood, and quality-of-life outcomes were similar in the 2 treatment groups (Table 3).

Our extended naturalistic follow-up of the CALM-PD trial participants provides insights into the frequency and types of antiparkinsonian treatments subjects are taking approximately 6 years after initiation of therapy with either levodopa or pramipexole. First, most patients in this cohort at 6 years from the decision to initiate dopaminergic therapy were receiving combination therapy regardless of the initial strategy. Second, more than 90% of subjects ended up receiving levodopa therapy regardless of the original treatment assignment. Third, more than 80% of those originally randomized to pramipexole were still taking a dopamine agonist (of those, 84.1% were taking pramipexole), although this does not account for the 43 subjects originally randomized to initial pramipexole who did not enter the CALM Cohort study and may not have been taking pramipexole.

These profiles may be an artifact of the CALM-PD study design whereby individuals randomized to initial levodopa treatment were unable to initiate dopamine agonist therapy during the initial 4 years of blinded follow-up. Subjects in the initial levodopa group were more likely to be receiving one of several other antiparkinsonian therapies (selegiline, anticholinergics, amantadine, entacapone), possibly to treat emerging dopaminergic motor complications that were more common in this group.²⁶ This may have reduced group differences in motor fluctuations or severity. Despite these differences in medication profiles, individuals in both groups tended to be receiving similar dosages of the various medications, suggesting that the initial treatment strategy and the resulting medication profiles at the time of assessment were more likely associated with the findings presented than the dosages.

The prevalence of dopaminergic motor complications (dyskinesias and wearing off) remained higher in the initial levodopa group, suggesting a long-term benefit on the development of these complications with initial pramipexole treatment. This finding was supported by a trend toward a greater percentage of time spent in the on state in the initial pramipexole group. However, motor complications did not have a clear effect on quality of life or disability, and while dyskinesias remained more common in the initial levodopa group, disabling dyskinesias were uncommon and no different between the treatment groups. It is possible that the greater use of amantadine in the initial levodopa group may have modified the severity of dyskinesias in this group.²⁸ Although changes from baseline in motor and ADL UPDRS scores remained lower (ie, better) in the initial levodopa group than in the initial pramipexole group, the differences were not as dramatic as those seen during the 4-year clinical trial, an attenuation that is most likely owing to medication crossover effects but may also result in part from bias due to individuals with more severe disease being lost to follow-up.

No apparent group differences in terms of self-reported disability, depressive symptoms, and quality-of-life scale scores were observed. In addition, varying methods for assessing disability revealed nearly identical results. Therefore, a more cumbersome method where subjects are asked to rate their disability in the on and off states that is weighted by the percentage of time spent in those states appears to be no more advantageous than the simpler approach where subjects rate their overall disability. This information may be useful in the assessment of disability in future long-duration longitudinal trials in PD.

There were significant differences between the 2 treatment groups with respect to the ESS scores and the prevalence of edema. Although we do not have baseline ESS scores (the ESS was not administered during the original trial), these differences persisted years after initiating therapy. This effect is likely owing to the greater proportion of subjects taking dopamine agonists at the final visit. While this discrepancy in dopamine agonist use is likely reflective of the initial treatment strategy, it is unclear when these group differences in the ESS became apparent over the 6 years of follow-up. Although sleepiness in PD is multifactorial, it appears unlikely that other unmeasured factors could have contributed to such a difference given that subjects in the 2 treatment groups had similar baseline characteristics, propensity score adjustment was performed, and all of the other outcome measures showed no group differences. In addition, this is consistent with a previous analysis of the 4-year CALM-PD trial that showed initial pramipexole treatment to be the most important risk factor for the development of somnolence as an adverse event.²⁹ We did not collect data on compulsive behaviors, which appear to be an increasingly worrisome complication possibly associated with dopamine agonist therapy.³⁰

Edema is a known complication of dopamine agonist therapy,^{29 ,31 - 33} but its functional effect is not well understood. This study suggests that despite an increased risk of edema in subjects initially treated with pramipexole, the edema had little effect on function or ambulation.

This study has several limitations. Despite the fact that 73.8% of patients enrolled in the CALM-PD trial entered the CALM Cohort study, individuals who did not enter the study were older, had more severe disease, and had worse quality of life. This suggests that the data presented here may reflect an overly optimistic snapshot of individuals approximately 6 years after initiating dopaminergic therapy. Although we adjusted for potential bias in the treatment group comparisons using propensity scores, residual systematic differences between the groups cannot be ruled out entirely. Finally, these results may not reflect actual practice in the community as all sites were academic movement disorders centers and the rigor of the initial study design may have influenced the final treatments and hence the clinical features of the illness.

These data complement recent reports comparing the long-term effect of the strategy of initial ropinirole vs initial levodopa⁸ and that of initial bromocriptine vs initial levodopa.⁹ In these reports, neither treatment strategy (dopamine agonist or levodopa) appeared to be superior in terms of its effect on disability or quality of life. The differences seen in favor of pramipexole (fewer dopaminergic events) or levodopa (less somnolence and edema) argue for an individualized approach. Over the long-term, there is no strong evidence favoring either of these initial treatment strategies over the other.

Correspondence:Kevin M. Biglan, MD, MPH, Department of Neurology, University of Rochester, 1351 Mt Hope Ave, Ste 220, Rochester, NY 14620 (kevin.biglan@ctcc.rochester.edu).

Accepted for Publication: October 3, 2008.

Published Online: March 9, 2009 (doi:10.1001/archneurol.2009.32).

Author Contributions:Study concept and design: Holloway, Marek, Biglan, Dick, Fahn, Kamp, Kieburtz, McDermott, Shinaman, and Shoulson. Acquisition of data: Marek, Julian-Baros, Kamp, Lang, Seibyl, Shinaman, Weiner, Pahwa, Grimes, Miyasaki, Johnston, Panisset, Factor, Evans, Shill, Harrigan, Hammerstad, Rajput, Jennings, Song, Fontaine, LeWitt, Wooten, Rost-Ruffner, Pfeiffer, Standaert, Tennis, Suchowersky, Pantella, Rodnitzky, Dobson, Kurlan, Berry, Kostyk, Riley, Jankovic, Atassi, Hunter, Waters, and Westerlund. Analysis and interpretation of data: Biglan, Dick, Kieburtz, Lang, McDermott, Shoulson, Weiner, Standaert, Suchowersky, Bennett, Brocht, Daigneault, Romer, and Watts. Drafting of the manuscript: Holloway, Biglan, Evans, Fontaine, Pantella, Dobson, Berry, Kostyk, and Atassi. Critical revision of the manuscript for important intellectual content: Marek, Biglan, Dick, Fahn, Julian-Baros, Kamp, Kieburtz, Lang, McDermott, Seibyl, Shinaman, Shoulson, Weiner, Pahwa, Grimes, Miyasaki, Johnston, Panisset, Factor, Shill, Harrigan, Hammerstad, Rajput, Jennings, Song, LeWitt, Wooten, Rost-Ruffner, Pfeiffer, Standaert, Tennis, Suchowersky, Rodnitzky, Kurlan, Riley, Jankovic, Hunter, Waters, Bennett, Brocht, Daigneault, Romer, Westerlund, and Watts. Statistical analysis: Dick, McDermott, Romer, and Watts. Obtained funding: Holloway, Marek, Kamp, Shinaman, and Shoulson. Administrative, technical, and material support: Holloway, Biglan, Fahn, Julian-Baros, Kamp, Kieburtz, Seibyl, Shinaman, Pahwa, Miyasaki, Panisset, Factor, Rajput, Fontaine, LeWitt, Rost-Ruffner, Pantella, Dobson, Bennett, Brocht, Daigneault, and Westerlund. Study supervision: Holloway, Kamp, Kieburtz, Shoulson, Weiner, Miyasaki, Wooten, and Kostyk.

Financial Disclosure: Dr Kieburtz has grant support from Boehringer Ingelheim for another study of pramipexole and has given expert testimony for Pfizer about pramipexole. Dr McDermott is currently the statistician for the following trial organized by the Parkinson Study Group and sponsored by Boehringer Ingelheim: “A Randomized, Double-Blind, Active- and Placebo-Controlled Efficacy Study of Pramipexole Given 0.5 mg and 0.75 mg BID Over a 12-Week Treatment Phase in Early Parkinson's Disease Patients (PramiBID).” This trial provided 20% support for Dr McDermott during the period of September 30, 2006, to September 29, 2008. Dr McDermott is currently consulting with Boehringer Ingelheim on another clinical trial involving pramipexole. In accordance with established conflict-of-interest policies of the Parkinson Study Group, he will receive no personal compensation for these consulting activities. Any compensation will be directed to the University of Rochester and deposited in a research account. Since this consulting arrangement was only recently established, no compensation has been received to date. Dr Pahwa is a consultant for Boehringer Ingelheim. Dr Pfeiffer has received royalties for books edited from Butterworth Heinemann (Elsevier), CRC Press (Taylor & Francis), and Humana Press, honoraria for lectures from GlaxoSmithKline, Boehringer Ingelheim, Novartis, Teva, Valeant, UCB/Schwarz, and Vernalis, honoraria for consulting from Boehringer Ingelheim, Prestwick, UCB/Schwarz, Vernalis, Kyowa, and Solvay, research grants or contracts from Kyowa, Novartis, Cephalon, Merck (Germany), Boehringer Ingelheim, Eisai, UCB/Schwarz, Santhera, and Solvay, and legal consulting fees from Spriggs & Hollingsworth and Davis, Graham & Stubbs. Dr Waters has received speaking honoraria from Boehringer Ingelheim.

Funding/Support:This study was supported by Pharmacia Corp and Boehringer Ingelheim.