Development and validation of a cefepime population pharmacokinetic model to guide treatment for patients with hospital-acquired pneumonia requiring renal replacement

ABSTRACT Cefepime is used in critically ill pneumonia patients, but dosing in renal replacement therapy (RRT) is complicated by altered renal function, extracorporeal removal, and variability in pulmonary epithelial lining fluid (ELF) exposure. We sought to develop an integrated model linking systemic, extracorporeal, and pulmonary cefepime pharmacokinetics in RRT. Pharmacokinetic (PK) data from critically ill adults receiving cefepime and requiring RRT were included. In a dedicated continuous RRT (CRRT) cohort ( n = 8), pre- and post-filter plasma and effluent PK data were collected. Additional patients ( n = 43) requiring RRT also contributed plasma and ELF samples. A non-parametric population PK model was developed using an 80/20 subject-level development-validation split. Covariate models assessed the impact of residual renal function, effluent flow rate, body size, and fixed intradialytic hemodialysis (HD) clearance. Monte Carlo simulations evaluated plasma and ELF target attainment across dosing strategies, weight strata, and minimum inhibitory concentrations (MICs). The final model jointly described systemic, extracorporeal, and pulmonary cefepime disposition using piecewise central volume, native clearance, linear effluent-flow-dependent CRRT clearance, fixed HD clearance, and ELF distribution. Simulations showed adequate plasma target attainment, whereas ELF attainment was sensitive to MIC, target, regimen, and body weight. Continuous infusions of 4 g/day provided adequate ELF attainment at MICs ≤4 mg/L. Plasma-based cefepime target attainment may overestimate pulmonary exposure during RRT, particularly at higher MICs, body weights, and f T>MIC targets. External validation and prospective studies are needed to refine pulmonary target estimates and determine whether plasma-guided Bayesian dosing can reliably ensure site-of-infection exposure.