Adenosine A2A receptor antagonist istradefylline (KW-6002) reduces “off” time in Parkinson's disease: A double-blind, randomized, multicenter clinical trial (6002-US-005)


  • Members of the 6002-US-005 Study Group are listed in the Appendix on page xx.



Based on new understanding of nondopaminergic pathways involved in Parkinson's disease (PD) pathophysiology, a selective adenosine A2A receptor antagonist, istradefylline, shows promise for the treatment of PD.


Istradefylline (40mg/day) was studied in levodopa-treated PD subjects experiencing prominent wearing-off motor fluctuations. At 23 North American sites, 196 subjects were randomized in a double-blind, 12-week outpatient clinical trial of istradefylline (114 completing the trial) or placebo (58 completing the trial). The primary efficacy measure was change from baseline to end point in the percentage of daily awake “off” time, recorded by subjects using a patient PD diary. Secondary end points evaluated “on” time (including “on time with dyskinesia”), the Unified Parkinson's Disease Rating Scale, and a Clinical Global Impression–Improvement of Illness score. Clinical laboratory, electrocardiograms, vital signs, and adverse event monitoring comprised the safety monitoring.


After randomization, approximately 88% of subjects completed the double-blind period. Compared with baseline, the decrease of daily awake “off” time for istradefylline was a mean (± standard deviation) of −10.8 ± 16.6% (95% confidence interval, −13.46 to −7.52) and for placebo, −4.0 ± 15.7% (95% confidence interval, −7.73–0.31; p = 0.007 using two-way analysis of variance). This effect corresponded to changes from baseline in total daily awake “off” time of −1.8 ± 2.8 hours for istradefylline and −0.6 ± 2.7 hours for placebo (p = 0.005). Treatment-emergent adverse effects with istradefylline were generally mild.


Istradefylline was safe, well tolerated, and offered a clinically meaningful reduction in “off” time without increased troublesome dyskinesia. Ann Neurol 2008

Although Parkinson's disease (PD) has several treatment options that initially can provide excellent symptomatic relief,1 control of its disabilities typically declines over time. Because PD is characterized by loss of dopaminergic neurons projecting from substantia nigra to striatal nuclei, the most rational and effective therapy for restoring dopaminergic neurotransmission has been the dopamine precursor L-dopa.2 Two years after starting L-dopa therapy, however, many patients start to experience fluctuations that interrupt control of parkinsonism, sometimes for up to several hours per day.3, 4 Adjusting the effects of L-dopa (by dosing changes or extenders such as catechol-O-methyltransferase or monoamine oxidase B inhibitors) or adding other dopaminergic drugs can improve “off” (undermedicated) states. Despite these options, inadequate control of motor fluctuations is a major source of disability for chronically treated PD.

Beyond restoring dopaminergic input to striatal neurons, other pharmacological interventions can influence response fluctuations and other motor features of PD. Potential targets are found in pathways that regulate neuronal signaling from the striatum to the globus pallidus externa and its accompanying projections onward to the subthalamic nucleus and substantia nigra pars reticulata. This circuit, the striatopallidal output pathway, is central to the pathophysiology of motor fluctuations and dyskinesias in PD.5 The striatopallidal output pathway arises in striatum predominantly from medium-sized spiny neurons that synthesize GABA and enkephalin as neurotransmitters, and express the A2A subtype of adenosine receptors.6 With pharmacological blockade of A2A adenosine receptors at the striatum and globus pallidus externa, there is enhanced striatal GABAergic inhibition on neurons in both regions and reduction of GABA release from striatopallidal terminals in the globus pallidus externa.7 The net result of adenosine A2A receptor antagonism is a reduction in the overactive striatopallidal output of the PD brain. This outcome, similar to that achieved from high-frequency electrical stimulation of the subthalamic nuclei, offers the potential for alleviating various motor impairments of PD.

Among several selective A2A adenosine receptor antagonists under development is a xanthine-based compound, istradefylline [previously designated as KW-6002; (E)-8-(3,4-dimethoxystyryl)-1,3-diethyl- 7-methyl-3,7-dihydro-1H-purine-2,6-dione]. Istradefylline has a 12nmol/ml binding affinity constant (Ki) in human brain.8, 9 Administered to neurotoxin-induced experimental models of parkinsonism in rodents10 and in nonhuman primates,9, 11 istradefylline augmented L-dopa in reversing motor impairments. Experiments conducted with dopamine D2 receptor–deficient mice showed that istradefylline acted independently of dopaminergic systems in counteracting impaired mobility.12 Furthermore, the adenosine A2A receptor specificity for the observed locomotor effects was confirmed by experiments in a nonhuman primate model of parkinsonism that showed reversal of the actions of istradefylline after administration of a selective A2A receptor agonist.13 In improving mobility for parkinsonian nonhuman primates, istradefylline treatment did not enhance L-dopa–induced dyskinesias.9, 11

The potential for istradefylline as a therapy for PD was demonstrated in exploratory clinical studies with oral doses up to 80mg/day.14, 15 These studies indicated that a daily dose of 40mg would be adequate for testing the concept of A2A receptor antagonism. Positron emission tomography studies in healthy human subjects receiving systemically administered, [11C]-labeled istradefylline showed more than 90% receptor occupancy in the striatum after administration of istradefylline doses greater than 5mg.16

Subjects and Methods


Study eligibility required diagnosis of idiopathic PD determined by the United Kingdom Parkinson's Disease Society criteria,17 Hoehn and Yahr scale18 severity of 2 to 4, L-dopa–responsiveness for at least 1 year (with a daily intake of ≥4 doses, or ≥3 doses/day if ≥2 were a sustained-release formulation), and wearing off of antiparkinsonian benefit (“off” time) lasting at least 2 hours per 24 hours as measured by an “on/off” subject home PD diary.19 Other PD medications were permitted but could not be altered ≤4 weeks before or during the study.


Subjects received either oral tablets of istradefylline, 40mg per day (Kyowa Pharmaceuticals, Princeton, NJ) or identically appearing placebo as assigned by 2:1 randomization. Istradefylline has a time-to-peak plasma concentration of 2 to 5 hours and a mean elimination half-life of 70 to 118 hours. Its clearance is primarily hepatic, metabolized predominantly via the CYP3A4 pathway in which it has relatively modest effects as a CYP3A4 inhibitor.


This 12-week, double-blind, placebo-controlled, randomized study was conducted at 23 investigational sites in the United States and Canada. An institutional review board or ethics committee at each site approved the protocol and its amendments. Each subject provided written informed consent. The registry for the 6002-US-005 study was number NCT00456586.

The primary efficacy end point was the change from baseline to end point (week 12 value or the last available postbaseline value) in the percentage of daily awake time spent in the “off” state, based on the 24-hour home PD diary. During each 30-minute interval, subjects recorded whether their predominant condition was asleep, “off,” or “on” (experiencing good control of parkinsonism). If dyskinesias also occurred during the “on” time, subjects differentiated the involuntary movement severity into “troublesome” or “non-troublesome” categories using prespecified criteria.19 Study participants underwent detailed instruction and needed ≥80% concordance with the investigator's simultaneous ratings during observations in the outpatient clinic setting. On 2 consecutive days during the 7-day period preceding the baseline visit and at weeks 2, 4, 8, and 12, subjects completed the 24-hour “on/off” motor assessment. They returned to the clinic at weeks 2, 4, 8, and 12 for safety and other study evaluations.

Key secondary end points included: (1) change from baseline to end point in the total hours of daily awake time spent in the “off” state, (2) change from baseline by study visit in the percentage and total hours of daily awake time spent in the “off” state, and (3) change from baseline to end point and by study visit in the percentage and total hours of daily awake time spent in “on state without dyskinesia” and “on state with dyskinesia” (with the latter divided into “on state with troublesome dyskinesia” and “on state with non-troublesome dyskinesia”). Another variable used in the analysis was derived from the sum of hours rated as “on state without dyskinesia” and “on state with non-troublesome dyskinesia”; these combination data were termed “on state without troublesome dyskinesia.” Other secondary end points included changes from baseline to end point and by study visits rated with the Unified Parkinson's Disease Rating Scale20 and Clinical Global Impression-Improvement (CGI-I) assessments. The Unified Parkinson's Disease Rating Scale Part III (Motor Examination) score was determined in a morning “off” state (after overnight omission of PD medications) at baseline, week 4, week 12, and in an “on” state (2 hours after first morning PD medication) at baseline and at all study visits. Clinical laboratory safety tests, electrocardiograms, vital signs, and physical and neurological examinations were assessed at screening and study visits, together with adverse event monitoring.

Statistical Methods

The sample sizes of the groups were based on the expected difference between groups in the percentage of “off” time from baseline to end point for the intent-to-treat (ITT) population. A total of 120 randomized subjects in the istradefylline group was planned for obtaining at least 100 subjects in the ITT population. For the placebo group, 60 randomized subjects were intended to yield at least 50 subjects in the ITT population. The sample size of each group was sufficient to provide 80% power for detecting statistical significance at the two-sided α level of 0.050 for treatment group differences ≥50% of the applicable standard deviation.

All efficacy variables (except CGI-I) were analyzed using a two-way analysis of variance approach with the ITT population (using last observation carried forward). This model contained terms for investigator and treatment. Treatment-by-investigator interaction for the primary efficacy variable, tested at the two-sided α = 0.150, was not significant (p = 0.34). The normality assumption for the primary efficacy variable, tested using Shapiro–Wilks test (α level = 0.05), was not significant (p = 0.54). Based on these results, the main effect model was accepted for the primary analysis.

For all secondary efficacy variables, only the main effect analysis of variance model (with terms for investigator and treatment) was used. The CGI-I was analyzed using the Cochran–Mantel–Haenszel test, stratified by investigator. Statistical tests on the primary efficacy variable were interpreted using an α level of 0.050. No adjustment for multiple testing was made. Other secondary variables were examined using descriptive analyses. Safety analyses were based on all subjects that received at least one dose of study medication.



Of 196 randomized subjects, 195 received study medication between April 23, 2002, and May 5, 2003 (Fig 1), and were included in the ITT population. The numbers of subjects analyzed in the primary analysis of istradefylline and placebo were 129 and 66 subjects, respectively. One subject randomized to istradefylline did not receive study drug. Descriptive statistical analysis of the groups showed a good balance between istradefylline and placebo treatment assignments, with no clinically relevant differences in baseline demographic or PD characteristics (Table 1). In the ITT population, most subjects also received a dopaminergic agonist; other PD medications were similar between the treatment groups, as was daily L-dopa intake. After randomization, almost 88% of subjects in both groups completed the 12-week study and study medication compliance was nearly complete.

Figure 1.

Subject flow diagram. Asterisks designate adverse events resulting in discontinuation were somnolence, asthma aggravated, pain in limb, muscle rigidity, dyskinesia (two subjects), dystonia, Parkinson's disease NOS, anxiety, sleep disorder NOS, chest pain, nausea (two subjects), muscle rigidity, tremor, bradykinesia, asthenia, motor dysfunction NOS, headache, and dizziness for istradefylline; and for placebo, adverse events resulting in discontinuation included balance impaired NOS, pneumonia NOS, bundle branch block, bradykinesia, abdominal pain NOS, and nausea. NOS = not otherwise specified.

Table 1. Demographics and Baseline Parkinson's Disease Characteristics (Intention-to-Treat Population)
inline image


The primary efficacy outcome variable showed a −10.8 ± 16.6% (95% confidence interval [CI], −13.46 to −7.52) change from baseline to end point in the percentage of daily awake time spent in the “off” state for istradefylline, and a placebo change of −4.0 ± 15.7% (95% CI, −7.73–0.31; p = 0.007 using two-way analysis of variance; Table 2). Expressed in hours, changes from baseline to end point in daily awake “off” time were −1.8 ± 2.8 hours (95% CI, −1.28–0.08) for istradefylline and −0.6 ± 2.7 hours (95% CI, −2.26 to −1.26) for placebo (p = 0.006; see Table 2). This corresponded to reductions in prior daily “off” time of 28% for istradefylline and 10% for placebo. The response to istradefylline occurred early (by the second week) and continued throughout the 12-week trial (Fig 2). Reductions from baseline in the percentage of daily awake time spent in an “off” state were significantly greater for istradefylline than placebo at weeks 2, 8, and 12, and exceeded placebo “off” time reduction by at least 1 hour (p ≤ 0.014).

Table 2. Efficacy Results: Mean Change (SD) from Baseline to End Point (Intent-to-Treat Population)
inline image
Figure 2.

Change from baseline to end point and by study visit in the percentage of daily awake time spent in the “off” state; observed-case analysis (intent-to-treat population) is shown. Circles represent placebo; triangles represent 40mg/day istradefylline. *Last observation carried forward. †p ≤ 0.05 versus placebo by analysis of variance.

The reductions in the percentage and total hours of daily awake time spent in the “off” state correlated with the increases in the percentage and total hours of daily awake time spent in the “on” state with and without dyskinesia. For the secondary variable “on time without dyskinesia,” istradefylline showed a small increase (0.17 hour) over placebo, although the treatment group difference was not significant. For “on time with dyskinesia,” a 1.00-hour increase (p = 0.035) occurred with istradefylline as compared with placebo. The “on time with dyskinesia” was predominantly rated as “non-troublesome.” Further analysis of the increase in “on” time demonstrated that the major component of the improvement of istradefylline over placebo was predominantly in “on time without troublesome dyskinesia” (ie, the combination of “on time without dyskinesia” and “on time with non-troublesome dyskinesia”), an increase of 0.96 hour over placebo (p = 0.026).

Other secondary efficacy analyses explored the Unified Parkinson's Disease Rating Scale components and total scores between baseline and end point; none showed consistent differences between istradefylline and placebo. At end point, 53.5% of subjects in the istradefylline and 40.9% of subjects in the placebo groups demonstrated improvement from baseline in CGI-I findings. The treatment group differences achieved statistical significance at weeks 4 and 8, but not at end point.

Adverse Events

The study treatment was generally well tolerated, and the incidence of all treatment-emergent adverse events was similar for both groups (istradefylline, 89.1%; placebo, 86.4%). Drug-related, treatment-emergent, adverse events (Table 3) were slightly greater for subjects receiving istradefylline (66.7%) than placebo (57.6%). Overall, the most frequently reported events were dyskinesia, dizziness, insomnia, nausea, and accidents involving falls. Of the most frequently reported drug-related events, dyskinesia occurred more often for istradefylline (30.2%) than for placebo subjects (15.2%), and accidents involving falls were more frequent for placebo (9.1%) than for istradefylline subjects (3.1%). Treatment-emergent dyskinesia was mostly mild or moderate in intensity. Severe dyskinesia was reported for two subjects in each group (istradefylline, 1.6%; placebo, 3.0%).

Table 3. Adverse Events Possibly or Probably Related to Study Drug Reported by 5% or More of Subjects in Either Treatment Group, by System Organ Class and Preferred Term (Safety Population)
inline image

Treatment-emergent serious adverse events were greater with istradefylline treatment (7.8%) than placebo (1.5%). Only one event reported on an istradefylline-treated subject was considered to be study drug–related (aggravated parkinsonism). One subject receiving placebo (1.5%) and two subjects receiving istradefylline (1.6%) died during the study; the investigators considered these deaths unrelated to study drug. The percentages of subjects discontinuing the study prematurely because of treatment-emergent adverse events were similar (7.0% and 7.6%) for both groups.

Mean changes from baseline over time for vital signs did not show any consistent differences between the two groups. The proportions of subjects with potentially clinically significant blood pressure changes were comparable for both groups. Four istradefylline (3.1%) and two placebo (3.0%) subjects had blood pressure decreases, and two subjects in each group had increases that met potentially clinically significant criteria. More istradefylline (7.0%) than placebo (3.0%) subjects had potentially clinically significant body weight changes. None of the changes in vital signs or body weight findings was regarded as clinically important. No clinically important differences were observed between the groups for clinical laboratory, electrocardiograms, or findings on physical or neurological examinations.


This study found istradefylline to be safe, well tolerated, and effective at improving end-of-dose wearing off in PD patients taking L-dopa. As compared with placebo, istradefylline provided approximately 18% reduction in daily “off” time (1.2 hours) over the course of the 12-week study in subjects, who, on average, were experiencing more than 6 hours of daily “off” time. Importantly, this reduction of “off” time occurred predominantly without increasing “troublesome” dyskinesia. Istradefylline provided these clinically meaningful benefits for PD subjects treated with conventional doses of L-dopa and who, in most instances, were also receiving adjunctive medications. In addition to L-dopa, 85.7% of the subjects were treated with a dopaminergic agonist and 41.0% were receiving entacapone. Improvements were evident in clinical ratings by 2 weeks after starting istradefylline and continued throughout the 12-week trial.

Prior studies with istradefylline provided evidence for its adjunctive benefits in L-dopa–treated PD subjects. In a proof-of-concept study with 11 subjects receiving istradefylline at 40 and 80mg/day, there was a substantial augmentation of antiparkinsonian effect when administered with an L-dopa dose that, by itself, was suboptimal.15 Another randomized, placebo-controlled clinical trial tested istradefylline doses from 5 to 40mg/day and measured a variety of clinical outcomes in 83 subjects with wearing-off and L-dopa–induced dyskinesias.14 In this study, all istradefylline doses led to reduced “off” time by a mean of 1.2 ± 0.3 hours, whereas placebo was associated with an increase of 0.5 ± 0.5 hour (p = 0.004) over the 12-week study. In contrast with results of this study, which found istradefylline caused an increase of dyskinesias, these were not enhanced in the study by Hauser and colleagues.14 study. Their exploratory investigation lacked a primary end point and was not designed with adequate statistical power for a conclusive answer to questions about the efficacy of the drug. It did, however, find a clear separation between placebo and istradefylline with respect to lessening “off” time. Our study's results regarding dyskinesias differ from the results of istradefylline testing in nonhuman primate MPTP-induced parkinsonism,9, 11, 13 which showed that this drug did not exacerbate pre-existing dyskinesias.

These results provide further confirmation that selective A2A adenosine receptor antagonism with istradefylline has promise for decreasing “off” time, a common problem of advanced PD. Our findings are consistent with preclinical studies suggesting that this therapeutic strategy, like deep-brain stimulation of the subthalamic nucleus,21 acts by decreasing striatopallidal output.6–9 It is unclear whether other properties associated with adenosine A2A antagonism in rodent experiments (such as increased striatal synthesis and release of dopamine,22 and enhanced movement reaction times and motor readiness reponses23) are involved. The improvement of wearing-off responses for advanced PD is only one of several unmet needs, which include effective treatments for “freezing” of gait and unpredictable “off” states. Whether istradefylline might also offer benefit for these problems remains to be evaluated. Another promising direction for future investigation with istradefylline is for neuroprotection against progression of PD, as suggested by experiments in mice with neurotoxin-induced parkinsonism.24


N.M.S. is an employee of Kyowa Pharmaceutical, Inc., and P.C.C. is a former employee.

A.M. is an employee of Kyowa Hakka Kogyo Co., Ltd. P.A.L. has received compensation for serving as a medical advisor to Kyowa Pharmaceutical, Inc.


This work was supported by Kyowa Pharmaceutical, Inc. (P.A.L., M.G., J.T., P.T. and the other investigators listed in the Appendix).

We acknowledge the assistance of Kyowa Pharmaceutical, Inc. in developing the research protocol, providing the data set, and conducting the statistical analysis of this clinical trial.


Dr P. LeWitt was the principal investigator for the 6002-US-005 study. The following investigators participated in the 6002-US-005 study: Drs A. Ahmed (Cleveland Clinic Foundation, Cleveland, OH); M. Guttman (The Centre for Movement Disorders, Ontario, Canada); B. Hiner (Marshfield Clinic, Marshfield, WI); D. Jennings (Molecular NeuroImaging, LLC, New Haven, CT); P. LeWitt (Henry Ford Health Systems, Southfield, MI); A. Lincoln (Agape Medical Research Center, Lubbock, TX); I. Litvan (University Neurologists, PSC, Louisville, KY); D. Meyer (Piedmont Medical Research, Winston-Salem, NC); P. O'Suilleabhain (University of Texas and Aston Ambulatory Care Center, Dallas, TX); S. Parashos (Struthers Parkinson's Center, Golden Valley, MN); A. Rajput (University of Saskatchewan, Saskatoon, Saskatchewan, Canada, and Wascana Rehabilitation Center, Regina, Saskatchewan, Canada); J. Rao (Louisiana State University, New Orleans, LA); D. Riley (University Hospitals of Cleveland, Cleveland, OH); B. Scott (Duke Movement Disorders Clinic, Durham, NC); K. Sethi (Medical College of Georgia, Augusta, GA); S. Sherman (University of Arizona Health Sciences Center, Tucson, AZ); C. Singer (University of Miami School of Medicine, Miami, FL); R. M. Stewart (Neurology Specialists of Dallas and Radiant Research, Dallas, TX); A. J. Stoessl (Pacific Parkinson's Research Centre, Vancouver, British Columbia, Canada); L. Struck (Iowa Health Physicians, Des Moines, IA); M. Swenson (University Neurologists, PSC, Louisville, KY); J. Tetrud (The Parkinson's Institute, Sunnyvale, CA); R. Trosch (Quest Research, Southfield, MI); P. Tuite (University of Minnesota, Minneapolis, MN); R. Watts (Emory University, Atlanta, GA); and R. Zweig (Louisiana State University, Shreveport, LA).