Impact of unbound fraction variations on piperacillin clearance and dosing recommendations
Within the dataset, patients were separated into four renal function categories based on their creatinine clearance (CL): (1) below 30 mL/min, (2) between 30 and 80 mL/min, (3) between 80 and 130 mL/min, and (4) above 130 mL/min. Afterwards, for each category, one patient with median renal function was selected for this evaluation. Thus, four patients with different profiles were retained. To simulate the effect of various unbound fractions of piperacillin in cases of hypoalbuminemia, we applied various factors to the total concentrations of piperacillin for all four subjects. As piperacillin’s unbound fraction is estimated at 70%, we started our evaluation with this value. A similar evaluation was performed for unbound fractions of 75, 80 and 85%. This resulted in four different unbound fraction scenarios, each containing four subjects for the assessment of target attainment.
A previously externally validated model was used to estimate piperacillin total CL for each patient within each data set through Bayesian estimation [15]. Individual CL estimation was realized by omitting the estimation step (MAXEVAL =0) and by fixing model parameters to the mean population estimates reported by the authors.
Once individual CL was obtained, dosing simulations were performed to determine whether unbound fraction variations influenced target attainment of piperacillin. Patient PK parameters were inputted on NONMEM, and the simulated piperacillin dose was based on the patient’s renal function; i.e., a loading dose of 4 g followed by a maintenance dose of 8, 12 or 16 g if CLCr was below 30, between 30 and 80, or above 80 mL/min, respectively, to reflect the practices reported by Klastrup et al. [11]. Simulations were repeated for the same subject with unbound fractions ranging from 70 to 85%. Target attainment was defined as 100% f T > MIC, for a MIC value of 16 mg/L, corresponding to the EUCAST clinical breakpoint of Pseudomonas aeruginosa . Concentration-time plots were generated to compare the various profiles for each subject.