Placebo and nocebo responses in gabapentin enacarbil and pregabalin trials for restless legs syndrome: time-course model-based network meta-analysis and meta-analysis of proportions.

PURPOSE: Placebo responses are often stronger in randomized controlled trials (RCT) on conditions that measure subjective improvement in symptoms, such as restless legs syndrome (RLS). This meta-analysis sought to quantify both placebo and nocebo responses in published RCTs of gabapentin enacarbil (GEn) and pregabalin. METHODS: In the first meta-analytic approach, we evaluated the time-courses taken by the participants randomized to the different doses of GEn and pregabalin and to placebo in their responses in International RLS Study Group Rating Scale (IRLS) scores. This was primarily evaluated by estimating the posterior median ET50s, the time (days) at which 50% of maximum response was achieved in IRLS scores for these participants. In a second meta-analytic approach, one-arm meta-analyses estimated proportions of participants on placebo considered “responders” on the investigator-initiated Clinical Global Impression-Improvement (CGI-I) scale, and separately of those who developed somnolence and dizziness. RESULTS: A total of 16 RCTs were included in this meta-analysis. Relative differences in ET50s (median and 95% credible intervals) between GEn 600, 900, 1200, 1800, 2400 mg/d doses and placebo were 1.09 (0.07; 2.92), 1.35 (0.05; 5.85), 0.56 (0.02; 1.79), 0.62 (0.03; 2.34) and 1.61 (0.10; 5.22) days, respectively, and between pregabalin mean doses of 107.8 and 337.5 mg/d and placebo were 21.75 (0.89; 145.26) and 2.54 (0.12; 8.41) days, respectively, indicating largely clinically insignificant differences. Pooled proportion of CGI-I responders amongst the participants randomized to placebo was 0.40 [(95% confidence intervals (CI): 0.31; 0.49), I2 = 84%]. Duration of RCTs and previous RLS treatment were associated with CGI-I response based on meta-regression analyses. Pooled proportions of participants randomized to placebo who developed somnolence and dizziness were 0.05 [(95%CIs: 0.03; 0.08), I2 = 75%], and 0.02 [(95%CIs: 0.01; 0.04), I2 = 62%], respectively. CONCLUSIONS: GEn and pregabalin showed substantial placebo responses in RCTs. Further studies are needed that would identify predictor variables for placebo responders or non-responders. REGISTRATION: OSF.IO/79JCF. REGISTRATION: The protocol was registered in OSF (OSF.IO/79JCF).