Population pharmacokinetic modelling of gentamicin and vancomycin in patients with unstable renal function following cardiothoracic surgery

PDF of poster

Objectives: To describe the population pharmacokinetics (PopPK) of gentamicin and vancomycin in cardiothoracic surgery patients with changing renal function and to evaluate whether new methodologies¹ for handling time-varying covariates offer advantages in data fitting.

Methods: Data were collected from cardiothoracic surgery patients. PopPK analysis was performed using NONMEM. Covariates investigated for an influence on clearance (CL) and volume of distribution (V) were sex, age, weight, ideal body weight, height, body surface area, days of therapy, serum creatinine and creatinine clearance (CrCL). Two new approaches to modelling CL in terms of CrCL were examined:

1. CrCL was separated into a baseline (B_CrCL) effect and difference (D_CrCL) from baseline effect (ie. CrCL – B_CrCL). P_CL = q _CL x [1 + q _BCrCL x (B_CrrCL – B_MedianCrCL) + q _DCrCL x D_CrCL].
2. The influence of CrCL on CL was allowed to vary between individuals by h ^CrCL,CLi. P_CL = q p x [1 + q _CrCL x exph ^CrCL,CLi (CrCL – CrCL_Median)] x exph ^Cli.

Predictive performances of final models were evaluated using independent data.

Results: Model building data comprised 96 patients (gentamicin) and 102 patients (vancomycin). Subjects were aged 17 to 87 years, with CrCL estimates of 9 to 172 ml/min. Changes in CrCL ranged from -76 to 58 ml/min (gentamicin) and -86 to 93 ml/min (vancomycin). Both data sets were adequately described by a linear relationship between CL and CrCL. Inclusion of B_CrCL and D_CrCL improved gentamicin modelling but had little impact on vancomycin modelling. Inclusion of h ^CrCL,CLi resulted in a poorly characterised model for gentamicin and no improvement in fit for vancomycin. Comparison of population and individual parameter estimates using independent data from 39 patients (gentamicin) and 37 patients (vancomycin) indicated no bias in CL. Mean (95% CI) differences were 4% (-3 to 11%) for gentamicin and 2% (-7 to 10%) for vancomycin. Final CL estimates were: CL_Gent(L/h) = 2.81 x (1 + 0.015 x (B_CrCL– B_CrCLMedian) + 0.0174 x D_CrCL. Interindividual variability (IIV) in CL was 27%. CL_Vanc(L/h)= 2.97 x (1 + 0.0205 x (CrCL – CrCL_Median). IIV in CL was 27%.

Conclusion: An additional parameter to describe individual changes in CrCL with time leads to improved PopPK modelling of gentamicin but not vancomycin in clinically unstable cardiothoracic patients.

Reference:
[1]. Wählby U, Thomson AH, Milligan PA, Karlsson MO. Models for time-varying covariates in population pharmacokinetic-pharmacodynamic analysis. Br J Clin Pharmacol 2004; 58(4): 367-377.