Model-based meta-analysis of the effects of non-selective and α1-selective GABAA receptor agonists in healthy volunteers

Purpose To quantify pharmacokinetic (PK) and pharmacodynamic (PD) relationships of various classes of GABA(A) agonists in healthy volunteers, in order to investigate the sensitivity of the biomarker responses due to differing GABA(A)-subtype selectivity and to explore the correlation between biomarker responses and side effects of these drugs. Methods A comprehensive search was conducted for published placebo-controlled clinical studies of non- and alpha(1)-selective GABA(A) drugs in healthy volunteers. PK/PD models were developed for concentrations and biomarker outcomes (saccadic eye movement (SEM), visual analogue scale (VAS), digit symbol substitution task (DSST), and critical flicker fusion test (CFFT)) extracted from included studies. Predicted responses and equivalent doses for biomarkers (based on predicted response) were used to compare drug effects. And the relationship between biomarkers and safety was explored by linear regression. Results A total of 2237 data from 163 articles were included. Based on PK and placebo effect modeling, linear biomarker-concentration relationships well fit the data. The alpha(1)-selective compounds had similar equivalent doses for VAS, DSST, and CFFT (4.7-6.7 mg), which were about three to seven times lower than that for SEM (14.4-35.5 mg), while such difference was less evident for non-selective drugs. DSST had the highest correlations with incidences of somnolence and dizziness. Conclusions The integral PK/PD models of GABA(A) agonists were established in healthy volunteers. SEM was identified as the most sensitive biomarker in differentiating GABA(A) receptor alpha(1) subtype selective compounds. The exploratory analysis implied that different relationships existed between the drug effects on biomarkers and the adverse event profiles in healthy volunteers.