Abstract
Background/Objectives: Guidelines recommend adjusting amikacin dosing based on patients' renal function. Nevertheless, for critically ill cancer patients, the renal function equations based on serum creatinine levels have low or no correlation with amikacin clearance. Considering this, using real-world data, we built an amikacin PBPK model to predict amikacin plasma concentrations in critically ill oncologic patients stratified by renal impairment levels. Further, the model was applied for dose stratification and individualization (digital twin strategy) in this population. Methods: In the Therapeutic Drug Monitoring (TDM) study, 368 amikacin pharmacokinetic analyses from 184 critically ill cancer patients were enrolled in three cohorts. A full-body PBPK model was developed using PK-Sim v. 11.3. Results: The final PBPK model accounted for two groups of critically ill cancer patients with mild (creatinine clearance; CLcr ≥ 60 mL/min) or severe (CLcr < 60 mL/min) renal dysfunction. In the dose stratification strategy, at the 7th dose, cancer patients with CLcr ≥ 60 mL/min under regimens 20 mg/kg (q24h); 25 mg/kg (q24h); 25 mg/kg (q48h); and 30 mg/kg (q72h) have probability of ≥69% of the patients achieving the efficacy target (AUC/MIC > 80, MIC of 4 mg/L), while cancer patients with CLcr < 60 mL/min under regimens 7.5 mg/kg (q24h); 15 mg/kg (q24h); 15 mg/kg (q48h); and 20 mg/kg (q36h) have ≥90% probability of achieving the same efficacy target. Conclusions: Our MIPD approach demonstrates potential in optimizing amikacin dosing for critically ill cancer patients. However, it does not eliminate the need for TDM due to unexplained variability still not accounted for by the PBPK model.