Eva Germovsek (1), Alison Kent (2), Nigel Klein (1), Paul Heath (2), Elisabet I Nielsen (3), Joseph F Standing (1)
(1) UCL Institute of Child Heath, Infectious Diseases and Microbiology Unit, London, UK; (2) St George’s, University of London, London, UK; (3) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: Gentamicin is an antibiotic with a narrow therapeutic window, making therapeutic drug monitoring (TDM) vital. However, currently there is no consensus on how best to perform gentamicin TDM in neonates. Previous studies have not found a relationship between serum creatinine concentration (SECR) and gentamicin clearance (CL), but these have often not accounted for the rapid changes in expected creatinine levels over the first weeks of life. This study aims to improve gentamicin TDM in neonates by firstly building a population PK (popPK) model and then using it to create a Bayesian computer tool, which will predict trough levels from levels taken at earlier and more convenient times.
Methods: The concentration-time data used for modelling were from two published studies [1, 2] and included 174 neonates and 1163 serum levels. Their gestational age ranged from 23.3-42.1 weeks, and postnatal age (PNA) from 0-65 days. To establish PK parameters NONMEM VII with FOCE-I method was used. In order to define the basic PK model several structural models and different error models were tested. The effect of different covariates (age, body weight, SECR) on the PK parameters was evaluated. SECR was incorporated into the model through organ function, which corrected measured SECR for age by using a SECR typical for that age from the literature [3, 4]. Inter-occasion variability (IOV) was tested for each parameter.
Results: A 3-compartment structural model described the data best. The final covariate model included allometric weight scaling of PK parameters, a hyperbolic sigmoid maturation function (with parameters fixed to values from a previous study [5]) and also a logistic function to explain the changes in CL with PNA. Age-corrected SECR was found to further significantly improve the model (the OFV decreased by 35.6 units). Final estimates for PK parameters were: CL=6.85 L/h/70kg, Vc=27.5 L/70kg, Q1=0.29 L/h/70kg, Vp1=157.1 L/70kg, Q2=2.24 L/h/70kg, Vp2=21.4 L/70kg. Between-subject variability (%CV) was: 25.7, 9.64, 14.5, 11.7 and 14.7 for CL, Vc, Vp1, Q2 and Vp2, respectively. IOV on CL was 15.6%. GOF and VPC plots showed a good fit of the model to the data and good predictive properties.
Conclusion: A popPK model for gentamicin in neonates that provided the best fit to the observed data was a 3-comparment model. We have shown that correcting SECR for age is necessary in neonatal studies.
References:
[1] Thomson et al., Dev Pharmacol Ther, 1988. 11(3): 173-9
[2] Nielsen et al., Clin Pharmacokinet, 2009. 48(4): 253-63
[3] Rudd et al., Arch Dis Child, 1983. 58(3): 212-5
[4] Cuzzolin et al., Pediatr Nephrol, 2006. 21(7): 931-8
[5] Rhodin et al., Pediatr Nephrol, 2009. 24(1): 67-76
Reference: PAGE 22 () Abstr 2914 [www.page-meeting.org/?abstract=2914]
Poster: Paediatrics