Bayesian Hierarchy model for population pharmacokinetics of amikacin in Japanese clinical population.

Amikacin is one of the aminoglycosides with a narrow therapeutic window, significant dose-response relationship, and substantial interindividual pharmacokinetics (PK) variability, thus requiring an individualized dosing regimen. In this paper, a three-stage Bayesian hierarchical model was developed based on the known the PK parameters of amikacin obtained from a nonlinear mixed-effects model established for the Japanese population, and the weights were assigned to the priori and posteriori parts before two-dimensional Gibbs sampling, and simulations were performed using the data of 24 elderly patients with respiratory tract infections in Japan, after analyzing the predicted values and the range between effective trough concentrations (Ctrough) and peak concentrations (Cpeak), and residual plots, the dose for patients 3, 7, 9, and 16 was increased to 600 mg/day, and the dose for patient 20 was decreased to 400 mg/day, while keeping the remaining patients' doses unchanged, and the serum concentration at the time of the last administration was predicted, which showed that the Bayesian hierarchical model and the Markov chain Monte Carlo algorithm in this study performed well.