192: Unraveling the Roadmap for Amikacin TDM in Critically Ill Cancer Patients

Statement of Purpose: Amikacin (AMK) is commonly used to treat gram-negative bacterial infections in ICU patients, but its narrow therapeutic range and high interindividual variability can lead to toxicity and ineffectiveness. For this reason, AMK therapeutic drug monitoring (TDM) is recommended. AMK irreversibly binds to the 30S bacterial ribosome subunit, inhibiting protein synthesis. It has a concentration-dependent effect and the best PK/PD index is a peak over MIC ratio of at least 8-10 to prevent resistance. Critically ill cancer patients may show high fluctuations in renal function biomarkers (current clinical marker to inform AMK dosing), challenging AMK TDM and dose selection in this special population. The goal of this study was to establish the roadmap for AMK TDM in critically ill cancer patients to finally provide a Bayesian estimator with bedside applicability (Figure 1). For that, we develop as a first step a population pharmacokinetic (PopPK) model to characterize individual PK parameters and identify major factors contributing to variability.

Description of Methods & Materials: An observational retrospective cohort study was carried out on oncologic patients admitted to the ICU. Patients were treated with AMK doses ranging from 5.76 to 39.12 mg/kg after 30-min IV infusion and under extended dosing intervals. Blood samples were collected in the peak and trough. Serum creatinine was evaluated at 0, 24, 72, 168 and 360 h after AMK dosing. Plasma concentration data of AMK were analyzed using a nonlinear mixed-effect modeling approach with NONMEM® v7.4. Sex, aged, body size descriptors (body weight, BMI, BSA), estimated glomerular filtration rate (eGFR), mechanical ventilation, concomitant drugs and comorbidities were tested as potential covariates.

Data & Results: A total of 246 concentration-time points from 51 patients were evaluated. Patients were mostly elderly, with sarcopenia, and altered renal clearance (including reduced or augmented and unstable renal function). The final model that described the concentration-time profiles was one compartment, linear elimination with eGFR (CKD_EPI equation, 2021) as a covariate on clearance and BSA on volume. The inclusion of eGFR explained 20% of the between-patient variability in clearance while BSA only accounted for 3% of the variability in volume. The model evaluation showed good predictive performance and stability. The individual model predictions identified 68% of the patient’s concentrations with less than 20-30% error supporting its potential application in the clinical setting. However, a more exquisite collection of patient information, including other biomarkers for renal function in ICU patients, could help to further refine the model and improve its predictive performance in this special population.

Interpretation, Conclusion or Significance: The PopPK model developed was able to unravel the roadmap to guide the AMK individualized dose adjustment in critically ill cancer patients. Next steps include a better characterization of the renal functionality in ICU cancer patients and evaluation of the probability target attainment based on the MIC regional frequency. The model will be then updated with the new information and use for developing a Bayesian estimator for bedside application.

Disclosures: All authors declared no conflicts of interest to the content of this abstract.