A Bayesian Framework for Optimizing Amikacin Therapy in Critically Ill Patients With Cancer

This is the authors' abstract. We don't add key points for this paper.

BACKGROUND: Amikacin (AMK) is used to treat gram-negative bacterial infections in intensive care unit (ICU) patients. However, its narrow therapeutic range and high interindividual variability can lead to toxicity and ineffectiveness. This study aimed to establish a roadmap for AMK therapeutic drug monitoring in critically ill patients with cancer to provide a Bayesian estimator of bedside applicability. METHODS: An observational retrospective study was conducted on oncological patients admitted to the ICU, treated with AMK as a 30-min intravenous infusion at 5.8-39.2 mg/kg. The plasma concentrations were analyzed using a nonlinear mixed-effects modeling approach. Covariate analyses were performed using anthropometric and laboratory data, concomitant drugs, and comorbidities. The model predictive performance was compared with previous AMK dosing approaches using the Bland-Altman method. RESULTS: The concentration-time profiles were best described using a one-compartment model with linear elimination. The estimated glomerular filtration rate was a significant covariate of clearance (CL), explaining 16% of the interpatient variability. Body weight was positively correlated with the volume of distribution, accounting for 4% of the variability. Our model reduced the bias in the estimates of individual CL values compared with that of other available methods and was further implemented in DoseMeRx for real-time application at the bedside. CONCLUSIONS: This study provides an effective example of a Bayesian estimation method for individualizing AMK doses in critically ill patients with cancer. Collecting more comprehensive patient information, including additional biomarkers for renal function, could further refine the model and improve its predictive performance in this special population.