Background and Objectives
The pharmacokinetics (PK) of piperacillin/tazobactam (PIP/TAZ) is highly variable across different patient populations and there are controversies regarding non-linear elimination as well as the fraction unbound of PIP (fUNB_PIP). This has led to a plethora of subgroup-specific models, increasing the risk of misusing published models when optimising dosing regimens. In this study, we aimed to develop a single model to simultaneously describe the PK of PIP/TAZ in diverse patient populations and evaluate the current dosing recommendations by predicting the PK/pharmacodynamics (PD) target attainment throughout life.
Methods
Population PK models were separately built for PIP and TAZ based on data from 13 studies in various patient populations. In the development of those single-drug models, postnatal age (PNA), postmenstrual age (PMA), total body weight (TBW), height, and serum creatinine (SCR) were tested as covariates. Subsequently, a combined population PK model was established and the correlations between the PK of PIP and TAZ were tested. Monte Carlo simulations were performed based on the final combined model to evaluate the current dosing recommendations.
Results
The final combined model for PIP/TAZ consisted of four compartments (two for each drug), with covariates including TBW, PMA, and SCR. For a 70-kg, 35-year-old patient with SCR of 0.83 mg L−1, the PIP values for V1, CL, V2 and Q2 were 10.4 L, 10.6 L h−1, 11.6 L and 15.2 L h−1, respectively, and the TAZ values were 10.5 L, 9.58 L h−1, 13.7 L and 16.8 L h−1, respectively. The CL for both drugs show maturation in early life, reaching 50% at 54.2 weeks PMA. With advancing age, CL of TAZ declines to 50% at 61.6 years PMA, whereas CL of PIP declines more slowly, reaching 50% at 89.1 years PMA. The fUNB_PIP was estimated as 64.5% and non-linear elimination was not supported by our data. The simulation results indicated considerable differences in PK/PD target attainment for different patient populations under current recommended dosing regimens.
Conclusions
We developed a combined population PK model for PIP/TAZ across a broad range of patients covering the extremes of patient characteristics. This model can be used as a robust a priori model for Bayesian forecasting to achieve individualised dosing. The simulations indicate that adjustments based on the allometric theory as well as maturation and decline of CL of PIP may help the current dosing recommendations to provide consistent target attainment across patient populations.
