Development of population pharmacokinetic models of total and free piperacillin in critically ill children and young adults for Monte Carlo simulations and model-informed precision dosing

Abstract Objectives We sought to develop population pharmacokinetic (PopPK) models of total and free piperacillin in critically ill children and young adults. We investigated the impact of kidney function and injury biomarkers and the effect of time on pharmacokinetic (PK) parameters. Methods We enrolled patients admitted to a paediatric ICU and prescribed piperacillin/tazobactam. We collected residual blood from clinical samples for total and free piperacillin concentrations. PopPK models were developed using nonlinear mixed effects modelling. Specific covariates of interest included time after antibiotic initiation, using a 48 h threshold, and urine neutrophil gelatinase-associated lipocalin (uNGAL), a tubular injury biomarker. Monte Carlo simulations investigated target attainment using different dosing regimens. Results Total (n = 695) and free concentrations (n = 639) from 139 patients were best described by two-compartment models. Creatinine-based estimated glomerular filtration rate (eGFR), uNGAL, and time >48 h after initial dosing were significant covariates of both total and free piperacillin clearance. Regarding central volume, albumin and cumulative fluid balance were significant covariates on total and free piperacillin, respectively. Simulations showed a common regimen of 100 mg/kg piperacillin/tazobactam every 6 h over a 30 min infusion is only sufficient for targeting bacteria with MICs of 2 mg/L or less in patients >30 kg with eGFR <150 mL/min/1.73 m2 or <30 kg with eGFR <90 mL/min/1.73 m2. Conclusion Using a large sample of critically ill children, we developed total and free piperacillin population PK models and newly identified uNGAL and time as significant covariates on piperacillin clearance. These models may be used for model-informed precision dosing after external validation.