Model-informed precision dosing of vancomycin in Chinese adult patients receiving renal replacement therapy: Systematic evaluation of published pharmacokinetic models and dosing regimen simulations

Abstract

Objectives

The pharmacokinetics of renally cleared vancomycin are significantly altered in critically ill patients undergoing renal replacement therapy (RRT), affecting the achievement of therapeutic targets. We evaluated the predictive performance of RRT patient-based PopPK models for model-informed precision dosing and subsequently simulated optimal dosing regimens for this population.

Methods

Six adult PopPK models were systematically identified and evaluated using a dataset of 226 concentrations from 23 adult patients on RRT from two study centers. Predictive performance was assessed using simulation and prediction-based diagnostics for a priori dosing based on patient characteristics and Bayesian dosing incorporating more than one measured trough concentration.

Results

The Oda model showed the best performance in prediction-based evaluation, with a median prediction error of –8.4%. Bayesian forecasting incorporating at least one measured trough concentration improved predictive capability, and the inclusion of dialysis filter characteristics in the model may further enhance its performance. For a 70 kg patient under this scenario: A 1 g loading dose during the first dialysis session achieves target concentrations within 24–48 h. Subsequent 0.5 g maintenance dose should be administered post-dialysis. Dose escalation is required for extended dialysis sessions (≥12 h).

Conclusion

The Oda model demonstrated the best predictive capability for dose prediction among the tested models in Chinese adult patients receiving RRT. However, its F20 and F30 values were undershot compared to the standard external evaluation criteria of F20 ≥35% and F30 ≥50%, highlighting the need for further study to optimize the model or collect additional data to validate its performance.