Randomized, double-blind, placebo-controlled study of XP13512/GSK1838262 in patients with RLS

Design.

This study (XenoPort, Inc. protocol #XP052, clinical trials.gov identifier NCT00298623; Patient Improvements in Vital Outcomes following Treatment in Restless Legs Syndrome I [PIVOT RLS-I]) was a randomized, double-blind, parallel-group comparison of XP13512 1,200 mg and placebo, conducted between March 2006 and February 2007 at 22 US centers (see the appendix for a list of investigators and institutions at which trials were performed).

Patient population.

Men and women aged 18 years or older with a diagnosis of moderate to severe primary RLS using International Restless Legs Syndrome Study Group diagnostic criteria¹ were recruited. Eligible patients had RLS symptoms ≥15 days during the month prior to screening (or, if on treatment, similar symptom frequency before treatment initiation) and symptoms on ≥4 nights during the 7-day baseline period. Prior RLS treatment was discontinued at least 2 weeks prior to baseline. Patients also had an International Restless Legs Scale (IRLS) total score ≥15 at the beginning and end of the baseline period.

To eliminate the potential to confound the assessment of treatment response, patients were excluded if they had evidence of secondary RLS; had a body mass index ≥34 kg/m²; were currently experiencing or being treated for moderate to severe depression, other primary sleep disorders, or neurologic disease or movement disorders; or had a history of RLS symptom augmentation or end-of-dose rebound with previous dopaminergic treatment. Pregnant patients were excluded. Although presence of daytime RLS symptoms (between 10:00 am and 6:00 pm) for ≥2 days during the week prior to baseline was originally an exclusion criterion, this restriction was removed after enrollment of approximately 30% of the total study population to improve the generalizability of study results.

All patients provided written informed consent prior to study participation. The study was conducted in accordance with good clinical practice guidelines and the 1996 version of the Declaration of Helsinki. The protocol was reviewed and approved by a local or regional institutional review board, depending upon center requirements.

Protocol.

Patients were randomized 1:1 to receive placebo or XP13512 1,200 mg once daily at 5:00 pm with food, using a blocked randomization schedule stratified by study site. Patients took one placebo or XP13512 600-mg extended-release tablet on days 1 to 3 and two placebo or 600-mg extended-release tablets on days 4 to 84. Eligible patients then entered an extension study or started a 7-day taper period (one placebo or XP13512 600-mg extended-release tablet). Blinding was ensured by the use of matching placebo and XP13512 tablets. If patients developed intolerable AEs during titration, the dose was maintained until the AE abated, decreased to the prior dose level, or withheld for a few days and then reinitiated to achieve the target dose. Clinic visits took place on days −7 and 1 (baseline), and at weeks 1, 2, 3, 4, 6, 8, 10, and 12.

Outcome measures.

Statistical analyses.

Statistical analysis was supervised by Daniel Bonzo, PhD, Senior Director and Head, Biometrics and Data Management, XenoPort, Inc. The sample size was determined for each of the two primary measures using the results of Phase 2 studies.²⁶ A sample size of 105 patients per treatment group was considered sufficient to detect, with overall 81% power, a treatment difference of −4.0 in mean change from baseline IRLS total score at the 0.05 significance level using a two-sided t test, assuming a SD of 8.8, and a response difference of 23% on the investigator-rated CGI-I (XP13512 68%, placebo 45%; odds ratio [OR] 2.6). Positive evidence of efficacy required both coprimary tests to be significant (p < 0.05).

All efficacy analyses were performed on the modified intent-to-treat population (MITT; all patients who took at least one dose of study medication and completed a baseline and at least one on-treatment IRLS assessment) using last observation carried forward (LOCF) as an imputation method. Coprimary efficacy endpoints were the mean change from baseline IRLS total score and the proportion of responders on the investigator-rated CGI-I at week 12 LOCF. The primary efficacy measures were also analyzed using observed cases at each time point, as secondary endpoints. Least squares (LS) mean treatment difference was calculated for the mean change from baseline IRLS total score using an analysis of covariance (ANCOVA) model adjusted for pooled site and baseline scores. Logistic regression, with treatment and pooled site as explanatory factors, was used to analyze the proportions of responders on investigator- and patient-rated CGI-I. Subgroup analysis of pain outcomes was performed on patients with a baseline RLS pain score ≥4. Median time of first onset of RLS symptoms was derived from 24-hour diary data. No adjustments for multiple comparisons were made for secondary efficacy endpoints.

As differences in study site sample sizes could affect the statistical power and generalizability of the overall study findings, study sites were pooled regionally to form six larger consolidated sites before performing analyses that included adjustment for site. Continuous efficacy variables were analyzed using an ANCOVA model with treatment and pooled site as main effects and baseline as a covariate. Dichotomous endpoints were analyzed by logistic regression using treatment and pooled site as explanatory factors.

Numbers needed to treat (NNT) for the coprimary endpoints were calculated post hoc using the smallest integer greater than or equal to one divided by the difference in response rates using LOCF. For IRLS total score, response was defined as a six-point decrease from baseline and a score <15. ESS was also analyzed post hoc using the same method described for the continuous efficacy variables.

Treatment compliance was defined as the ratio between the actual number of pills taken and the expected number of pills taken during the study period. Patients were considered compliant if the ratio was 0.8 to 1.2.

Patient disposition.

In total, 222 patients were randomized to receive XP13512 1,200 mg (n = 114) or placebo (n = 108) (figure e-1 on the Neurology® Web site at www.neurology.org). Two patients (XP13512) withdrew prematurely and were excluded from the MITT population. One had a conflicting work schedule, did not receive study drug, and was also excluded from the safety population. The other developed somnolence and withdrew prior to the week 1 assessment. Study completion rates were similar in the two groups.

Patient characteristics.

Baseline demographics, RLS history, and IRLS scores were similar across treatment groups (table 1) and reflected moderate to severe disease severity. The majority of patients (68.3%) were treatment naïve.

Compliance rates were 95.4% (XP13512) and 85.8% (placebo). The mean number of days on study drug was similar with XP13512 (77.6 days) and placebo (76.2 days).

Outcome measures.

Primary efficacy outcomes.

The mean (SD) change from baseline IRLS total score at week 12 (LOCF) was greater with XP13512 (−13.2 [9.21]) compared with placebo (−8.8 [8.63]) (adjusted mean treatment difference: −4.0; 95% CI, −6.2 to −1.9; p = 0.0003) (figure 1). More patients treated with XP13512 (76.1%) were rated by investigators as responders at week 12 (LOCF) compared with placebo (38.9%) (adjusted OR: 5.1; 95% CI, 2.8 to 9.2; p < 0.0001) (figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Mean (±2 SE) change from baseline International Restless Legs Scale (IRLS) total score by visit

Coprimary endpoint: adjusted mean treatment difference (week 12, last observation carried forward [LOCF]): −4.0 (analysis of covariance [ANCOVA]: 95% confidence interval: −6.2 to −1.9; p = 0.0003). ***p < 0.0001 vs placebo, **p = 0.0001 vs placebo, ^†p = 0.0003 vs placebo; ANCOVA with baseline as covariate and treatment and pooled site as main effects. Unadjusted mean change from baseline at week 12 (LOCF): XP13512 = −13.2; placebo = −8.8.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Investigator-rated responders on the Clinical Global Impression–Improvement (CGI-I) by visit (last observation carried forward)

Coprimary endpoint (week 12, last observation carried forward): OR: 5.1; 95% confidence interval: 2.8 to 9.2; p < 0.0001. ***p < 0.0001 vs placebo. Logistic regression adjusted for pooled site. Responders defined as patients with an investigator rating of much improved or very much improved on the CGI-I.

The post hoc NNT to gain one response benefit was 6 (95% CI, 3.1 to 14.3) for the IRLS total score, 3 (95% CI, 2.0 to 4.0) for the investigator-rated CGI-I, and 4 (95% CI, 2.4 to 6.5) when both primary endpoints were considered.

Secondary efficacy outcomes.

At week 1, the mean [SD] change from baseline IRLS total score was greater with XP13512 (−10.7 [8.01]) compared with placebo (−4.4 [7.21]) (adjusted treatment difference: −6.0; 95% CI, −7.9 to −4.1; p < 0.0001). Similarly, at week 1, more patients treated with XP13512 (57.9%) were rated by investigators as responders compared with placebo (24.8%; p < 0.0001). At week 12, more patients treated with XP13512 (73.6%) rated themselves as responders compared with placebo (42.6%; p < 0.0001). Similar results were reported at week 1 (61.7% vs 24.0%; p < 0.0001).

XP13512 increased the average daily TST at weeks 2 (LS treatment difference 0.3; p = 0.0033) and 4 (LS treatment difference 0.2; p = 0.0246), but not at week 12 (LS treatment difference 0.2; p = 0.1870) compared with placebo. Mean (SD) change from baseline average daily WASO was greater at week 12 with XP13512 (−17.6 [29.28]) compared with placebo (−11.8 [24.91]; LS treatment difference −6.5; p = 0.0033).

Mean reduction in average daily RLS pain score in patients with baseline RLS pain score ≥4 was greater at week 12 with XP13512 (−3.7) compared with placebo (−1.9) (LS treatment difference −1.7; 95% CI, −2.6 to −0.9; p < 0.0001).

XP13512 improved RLSQoL scores at week 12 compared with placebo (mean [SD] change from baseline: XP13512, 21.4 [17.00]; placebo, 14.1 [17.32]; LS treatment difference 7.8; p < 0.0001).

All MOS sleep scale domains improved with XP13512 at week 12 compared with placebo. XP13512-treated patients had greater mean [SD] changes from baseline to week 12 compared with placebo-treated patients in daytime somnolence (−17.4 [19.39] vs −9.6 [17.18]; p = 0.0018), sleep quantity (0.8 [1.24] vs 0.4 [1.38]; p = 0.0084), sleep adequacy (27.7 [29.89] vs 13.4 [27.42]; p < 0.0001), and sleep disturbance (−29.1 [23.84] vs −15.5 [21.79]; p < 0.0001). All PSQ sleep outcomes improved with XP13512 at week 12 compared with placebo (sleep quality p < 0.0001, next day functioning p = 0.0002, number of nights with RLS symptoms p < 0.0001, number of nighttime awakenings from RLS symptoms p = 0.0429, and number of hours awake due to RLS symptoms p = 0.0189).

At baseline, the median time to onset of first RLS symptoms based on the 24-hour patient diary was 6.0 hours in both treatment groups, and almost all patients exhibited symptoms by the end of the 24-hour period (figure 3). XP13512 increased the median time to onset of first symptoms to 13.3 hours at week 2 and to >23.5 hours at week 12 compared with placebo (9.0 hours at week 2, p = 0.0006; 11.5 hours at week 12, p < 0.0001). At week 12, 50.5% of patients treated with XP13512 were symptom-free at 24 hours compared with placebo (17.7%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Time to onset of restless legs syndrome symptoms (over 24 hours beginning at 8:00 am) at baseline and at week 12

Time to onset of first restless legs syndrome symptoms analyzed using the Kaplan-Meier method; log-rank test used to compare treatments.

Investigators and institutions at which trials were performed: Donald Ayres, MD (Ayres & Assoc. Clinical Trials, Lebanon, NH); Philip M. Becker, MD (Sleep Medicine Associates of Texas, P.A., Dallas); Eileen Brady, MD (Lovelace Scientific Resources, Albuquerque, NM); David Chen, MD (Beacon Clinical Research, LLC, Brockton, MD); John Cochran, MD (Alexandria Fairfax Neurology, PC, VA); Aaron Ellenbogen, DO, MPH (Quest Research Institute, Bingham Farms, MI); William Ellison, MD (Radiant Research, Greer, SC); Ramedevi Gourineni, MD (Northwestern Medical Faculty Foundation, Chicago, IL); Dennis Hill, MD (Central Carolina Neurology & Sleep, Salisbury, NC); John Hudson, MD (Future Search Trials, Austin, TX); David Kudrow, MD (Neurological Research Institute, Santa Monica, CA); Clete A. Kushida, MD, PhD (Stanford Center for Human Research, Palo Alto, CA); Antoinette Pragalos, MD (Community Research & Sleep Management Institute, Cincinnati, OH); Marc Raphaelson, MD, PA (Community Clinical Trials, Frederick, MD); Albert Razzetti, MD (University Clinical Research Deland, Inc., Deland, FL); Paul Scheinberg, MD (Atlanta Pulmonary Group–Research, GA); Markus Schmidt, MD, PhD (Ohio Sleep Medicine & Neuroscience Institute, Inc., Dublin, OH); Susan Steen, MD (Axiom Clinical Research of Florida, Tampa, FL); Stephen G. Thein, PhD (Pacific Research Network, San Diego, CA); Alberto Vasquez, MD (Suncoast Neuroscience Assoc, Inc., St. Petersburg, FL); Jan Westerman, MD (Jasper Summit Research, LLC, Jasper, AL); David Winslow, MD (Kentucky Research Group, Louisville, KY).