Prior-informed population pharmacokinetic-pharmacodynamic modeling of dexamethasone in horses

Dexamethasone (DEX) is widely used in equine practice with diverse studies having characterized its pharmacokinetics (PK) and pharmacodynamics (PD) including our published comprehensive meta-analysis. Individual DEX concentrations in plasma and urine and serum cortisol from five studies (n = 32 horses) were analyzed using a minimal physiologically-based PK (mPBPK) model and a circadian baseline-indirect response PD model. The previous meta-analysis provided complex structural models to assess sources of variability across an array of studies involving intravenous (IV), intramuscular (IM) and intra-articular (IA) dosing of various formulations. Here we compare three population estimation approaches: first-order conditional estimation with interaction (FOCEI), FOCEI with informative priors (FOCEI-Priors), and full Bayesian estimation. The primary covariate found was hepatic clearance of DEX being 17% higher in adult female horses (0.59 L/h/kg) than in geldings (0.49 L/h/kg). Between-subject variability was identified for some parameters. Standard FOCEI required extensive fixing of parameters to achieve convergence. FOCEI-Priors exhibited robust estimation performance, resulting in well-captured DEX PK and cortisol profiles, precise parameter estimates, and satisfactory model diagnostics. Bayesian estimation failed in the posterior exploration for cortisol PD, likely due to its greater sensitivity to prior specification and data variability. This work demonstrated effective population extensions of complex, prior-informed, mechanistic mPBPK/PD models to confirm determinants and assess variability of DEX disposition and adrenal effects in horses.