Population pharmacokinetic model of gentamicin in paediatric oncology patients.

Aims: To describe the population pharmacokinetics of gentamicin in paediatric febrile neutropenic patients and to identify covariates which explain pharmacokinetic variability in this population.

 

Methods: Data for pharmacokinetic analysis were collected retrospectively from the medical records of paediatric febrile neutropenic patients who had undergone gentamicin concentration-time profiling at the Lady Cilento Children's hospital between 2008 and 2013. A population pharmacokinetic model of gentamicin was developed using a nonlinear mixed effects modelling approach with NONMEM® software version 7.3 with an Intel FORTRAN compiler and Perl Speaks NONMEM (PSN) version 3.5.3. Typical population pharmacokinetic parameters of gentamicin as well as between-subject and between-occasion variability were estimated in these patients via the LAPLACIAN with interaction estimation method. Xpose (v 4.0) and R software was used for data exploration and visualisation. One-, two- and three-compartment disposition models with first-order elimination were tested during model building. Proportional, additive and combined error models were compared to describe residual variability. Covariates screened to assess their influence on the pharmacokinetics of gentamicin included: post-natal age (PNA), body weight (WT), fat-free mass (FFM) (1), serum creatinine (Scr), renal function (eGFR) (2, 3) and concomitant nephrotoxic medication such as metrothexate, cyclophosphamide, ambisome, cyclosporine, cisplatin and carboplatin. Potentially influencing covariates were included in the model following a sequential forward selection and backward elimination step. patients. Gentamicin drug concentrations below the assay lower limit of quantification (LLOQ) were handled using method M3 (4). Difference in objective function value (ΔOFV), visual comparison with goodness-of-fit plots and also difference in magnitude of unexplained and explained parameter variability (5) was used for model comparison. Model predictive performance was evaluated via prediction-corrected PVC (pcVPC) based on 1000 simulations.

 

Results: Data were collected from 423 oncology paediatric patients aged 0.21-18 years (median: 5.18 years). Gentamicin disposition was well described by a two-compartment model with first order elimination. The best base model allowed estimation of between subject variability (BSV) associated with clearance (CL), volume of the central (V1) and peripheral (V2) compartments; correlation between CL and V1 and estimation of between occasion variability (BOV) in CL. Residual unexplained variability (RUV) was better described using a combined error model. A model that included a covariate effect of WT on CL, V1, Q and V2 and PNA on CL had increased explained parameter variability on CL. Parameter estimates (BSV (%CV)) were CL (L/h/70kg): 5.75 (18.2%), V1 (L/70kg): 16.2 (22.2%), Q (L/h/70kg): 0.86 and V2 (L/70kg): 4.36 (65.3%). BOV (%CV) on CL was 15.5%. The proportional error component of RUV (%CV) was 35.3% and the additive error component was 0.04 mg/L. 

 

Conclusions: PNA and Scr influence the pharmacokinetics of gentamicin in oncology paediatric patients. These covariates provided a better fit and increased explained variability on CL. The effects of concomitant nephrotoxic medication on the pharmacokinetics of gentamicin are still to be investigated.

 

References:

1. Al-Sallami HS, Goulding A, Grant A, Taylor R, Holford N, Duffull SB. Prediction of Fat-Free Mass in Children. Clinical pharmacokinetics. 2015;54(11):1169-78.

2. Rhodin MM, Anderson BJ, Peters AM, Coulthard MG, Wilkins B, Cole M, et al. Human renal function maturation: a quantitative description using weight and postmenstrual age. Pediatric nephrology. 2009;24(1):67-76.

3. Schwartz GJ, Munoz A, Schneider MF, Mak RH, Kaskel F, Warady BA, et al. New equations to estimate GFR in children with CKD. Journal of the American Society of Nephrology : JASN. 2009;20(3):629-37.

4. Beal SL. Ways to fit a PK model with some data below the quantification limit. J Pharmacokinet Pharmacodyn. 2001 Oct;28(5):481-504.

5. Hennig S, Karlsson MO. Concordance between criteria for covariate model building. Journal of pharmacokinetics and pharmacodynamics. 2014;41(2):109-25.