Individualized Cefepime Dosing Study

Patients will received cefepime administered per standard-of-care as a 30 min intravenous infusion. Dosing will be based on local guidelines (the Sanford guide to antimicrobial therapy 2012-2013) using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine formula to estimate glomerular filtration rate (GFR).

Blood will be sampled immediately prior to dose administration (time = 0 at the start of the 30 min infusion), at 0.5, 1, 3, 5 hours post-start of infusion and just before the subsequent dose. From day two onwards, samples will be taken at the end of the infusion and just before the next dose. For the quantification of cefepime, a validated solid phase extraction-liquid chromatography electrospray-tandem mass spectrometry method will be used.

Timed urine collections were taken during one dosing interval (8 hours in a three times daily regimen) every day.

Creatinine (modified Jaffe method) and urea in serum will be determined using an Architect c16000 analyzer (Abbott, Chicago, IL, USA). Cystatin C will be determined using a particle-enhanced immunonephelometric assay (N Latex Cystatin C, Siemens Healthcare Diagnostics, Marburg, Germany) by use of a BN II nephelometer (Siemens Healthcare Diagnostics). This assay has a calibration traceable to the first certified reference material for cystatin C in human serum (ERM-DA471/IFCC). Kidney injury molecule-1 (KIM-1) in urine and uromodulin in serum will be determined using commercially available ELISA assays: Quantikine ELISA Human TIM-1/KIM-1/HAVCR (R\&D Systems, Minneapolis, MN, USA) and Uromodulin ELISA (Euroimmun, Luebeck, Germany), respectively.

The cefepime concentration versus time data will be fitted using the FOCE-I estimation algorithm in NONMEM® (Version 7.3; GloboMax LLC, Hanover, MD, USA). R® (R foundation for statistical computing, Vienna, Austria) will be used to graphically assess the model's goodness-of-fit and to evaluate the model's predictive capabilities. As a measure of prediction error, the absolute prediction error (APE) will be used. In short, the measured cefepime concentrations for each individual i at time point j were compared against the population predicted cefepime concentrations, i.e. the predictions for each individual without taking into account the between-subject variability (PRED in NONMEM). The distribution of APEs will be summarized by the median and 90% percentile.

Renal function will be assessed by four serum based kidney markers (serum creatinine, cystatin C, urea and uromodulin) and two urinary markers (measured creatinine clearance (CrCl) and KIM-1, both on timed urine collections). Serum creatinine and cystatin C will also be used to calculate the eGFR based on CKD-EPI formulas.

Based on the final covariate model, a Monte Carlo-based simulation study will be performed to evaluate the Sanford dose recommendations for ICU patients.