1402. Cystatin C Improves Estimation of Vancomycin Clearance in Critically Ill Children Using a Population Pharmacokinetic Modeling Approach

Background Vancomycin (VAN) is renally eliminated and clearance (CL) correlates with glomerular filtration rate (GFR). The bedside Schwartz equation (Schwartzbed), currently used to estimate GFR in children, relies solely on creatinine (Cr) and is inaccurate in critical illness. We compared the Schwartzbed against various GFR-estimating equations that incorporate the novel biomarker cystatin C (CysC) in a population pharmacokinetic (PK) model of VAN CL in critically ill children. Methods Children 2–18 years of age receiving intravenous VAN in the Children’s Hospital of Philadelphia PICU were enrolled. Three PK samples were collected during a single steady-state dosing interval in addition to VAN concentrations collected for clinical care. A sample was obtained prior to and during PK sampling for the measurement of CysC and Cr. VAN concentrations, dosing histories, and covariates (age, height, weight, sex, eGFR) were analyzed using nonlinear mixed-effects modeling with NONMEM v7.4. Model evaluation/selection was based on successful convergence, precision of the parameter estimates, the Akaike Information Criteria (AIC), and comparison of goodness-of-fit diagnostic plots of models including Schwartzbedand other published Cr- and CysC-based eGFR equations. Results We enrolled 20 subjects age 12.7 years (range: 3.9–18.2); six were female. Median VAN dosing at PK sampling was 57.4 mg/kg/day (range: 26.4–80.1). Median Cr was 0.35 mg/dL (IQR 0.3–0.5) and CysC was 0.5 mg/L (IQR 0.4–0.8); correlation between Cr and CysC was poor (0.24). Population PK data were described by a two-compartment model with allometric scaling for all parameters. The full age spectrum equation using both Cr and CysC [eGFR = 107.3/((Cr/QCr)*0.5 + (CysC/QCysC)*0.5); QCr and QCysC are normal values for age] as a covariate on CL had the largest reduction in AIC compared with Schwartzbed (ΔAIC −11.571) and provided best model fit. Typical population PK parameters (95% CI) normalized to 70 kg were 0.13 L/minute (0.11,0.14), 24.5 L (7.7,41.5), and 0.14 L/minute (0.01, 0.28) for CL, V1, and Q, respectively. Conclusion Incorporation of eGFR calculated using the Cr- and CysC-based full age spectrum equation improved population PK model fit for VAN among critically ill children compared with Schwartzbed. Clinical use of CysC may help estimate VAN CL among critically ill children compared with use of Cr alone. Disclosures K. Downes, Merck, Inc.: Investigator, Research support. Pfizer, Inc.: Investigator, Research support.