Population pharmacokinetic-pharmacodynamic (PK-PD) modeling was used to explore covariate effects on zolpidem PKs and PDs in 30 healthy Korean volunteers (15 males, 15 females) who received a single 10 mg dose. The dataset included 325 PK observations and 388 observations each for the digit symbol substitution test (DSST), choice reaction time (CRT), and visual analog scale (VAS). A one-compartment PK model with transit compartment absorption and first-order elimination was developed, followed by sequential sigmoid Emax-based direct response models for each PD endpoint. A hyperbolic learning effect model was incorporated for DSST, whereas CRT showed no clear learning effect. Baseline VAS was described using linear splines. No clinically meaningful covariates were retained in the final PK or PK-VAS models. In contrast, higher body weight was associated with reduced DSST IC50 and CRT EC50. Lower albumin was associated with increased potency for both DSST and CRT and with a steeper concentration-response relationship for CRT. Higher baseline CRT values were linked to greater potency and a steeper slope for CRT, and older age was associated with lower baseline DSST and CRT performance. These exploratory findings suggest that variability in zolpidem response may be partially mediated by physiologically and functionally relevant factors, such as body weight, albumin, age, and baseline PD measures, which could complement sex in individualized risk assessment.
Kim et al. (Sun,) studied this question.