A literature-informed pharmacokinetic framework for polymyxin B dosing during continuous renal replacement therapy: integrating renal function and extracorporeal clearance

Introduction Existing population pharmacokinetic (PopPK) models for polymyxin B in patients receiving CRRT remain limited, and all of them have neglected the impact of membrane materials on drug adsorption. Recently, a few studies have attempted to construct integrated PK models by combining pharmacokinetic (PK) parameters derived from patients not receiving CRRT with CRRT‐specific parameters. However, current integrated models only consider non‐renal clearance and CRRT clearance (CLCRRT), excluding residual renal function (RRF). This study aimed to construct a literature-informed PK framework for polymyxin B incorporating intrinsic clearance (CLintrinsic, adjusted by CrCL to account for RRF) and CLCRRT, with CLCRRT comprising transmembrane (CLTM) and adsorptive (CLads) components. Methods CLCRRT allocation was set to 0% CLads for the PS membrane and 100% CLads for the AN69 membrane. The frameworks predictive performance was evaluated using two literature datasets (n = 53 and n = 30) and a prospective cohort (n = 10). Simulation analyses used PTA ≥90% as the criterion for adequate pharmacodynamic target attainment and a simulated probability of AUC 0‐24 h > 100 mg·h/L ≤ 14.5% as the exposure‐exceedance threshold. Results The framework demonstrated accurate predictions of polymyxin B exposure in patients not receiving CRRT and those receiving CRRT (prediction error <30%). Filter type was the primary determinant of drug exposure, exceeding the effects of RRF or CRRT intensity. For the PS membrane (AV600S filter) and MIC = 0.5 mg/L pathogens, doses should be stratified based on CrCL with standard or lower doses. For the AN69 membrane (OXIRIS filter), a high‐dose regimen (200 mg LD + 150 mg MD q12h) should be considered for infections with an MIC of 0.5 mg/L. An interactive Shiny application ( https://xy3yx-cn.shinyapps.io/polymyxin-B-crrt-dosing/ ) was developed. Discussion These findings suggest that filter type and RRF should be considered when individualizing polymyxin B dosing during CRRT. The proposed framework and the Shiny application are intended as research tools and require prospective clinical evaluation and therapeutic drug monitoring before routine clinical use.