Improving gentamicin dosing in paediatric oncology patients

Objective: To develop and apply a population PK model of gentamicin in paediatric oncology patients in i) a utility function approach that balances the probability of efficacy against potential kidney damage related to gentamicin accumulation in the renal cortex and ii) a semi-mechanistic modelling approach to simulate bacterial killing over time under different doses; with the ultimate goal of determining the optimal initial dose of gentamicin in this population.

Methods: Data were collected retrospectively from paediatric oncology patients who received gentamicin between 2008 and 2015. NONMEM® was used to develop a population PK model and estimate an optimal initial dose using a utility function. The model was developed using physiological plausible covariate parameterization. In the utility function, probability of efficacy was defined based on efficacies reported previously for different exposure targets: Cmax/MIC=10 and AUC24= 100 (mg*h/L) or AUC24/MIC =100 (1, 2). Gentamicin accumulation in the renal cortex, which may lead to nephrotoxicity, was calculated based on cumulative gentamicin uptake into renal cortical cells given an individual’s exposure divided by the maximal possible accumulation within a dosing interval (3). Bacterial kill under the optimal initial gentamicin doses estimated, given different microorganism MICs, were tested in two semi-mechanistic models (4, 5).

Results: Data from 423 patients (median body weight: 19.4 kg and median age: 5.2 years) receiving an average initial 30-minute gentamicin infusion of 7.0 mg/kg, were included in model development. Comparable data from 52 patients were used for model evaluation. A two-compartment model best described the pharmacokinetic data. Renal function maturation (6) and serum creatinine had a significant influence on gentamicin CL, whereas FFM influenced V1, Q and V2. Typical PK parameter estimates were: CL (L/h/70kg) = 5.77; V1 (L/70kg) = 21.6; Q (L/h/70kg) = 0.62 and V2 (L/70kg) = 13.8. Based on the utility function, the optimal initial dose ranged from 7.0 (MIC=0.5) to 10.5 mg/kg (MIC=4).  An initial dose of 8.3 mg/kg/24 hours was estimated to have on average 85% probability of achieving the defined Cmax/MIC and AUC24 targets (MIC = 0.5, 1 and 2 mg/L), at the same time 15% of the maximal possible gentamicin accumulation in the renal cortex was noted. Under that dose all patients had a concentration < 1 mg/L at 24 hours post-dose. When tested in the semi-mechanistic model this dose produced rapid initial bacterial killing with no bacterial regrowth for at least 12 hours for the typical patient and bacterial count did not reach the starting initial inocula at 24 hours. A dose of 7.0 mg/kg/24h, commonly administered in clinical practice, would not achieve adequate efficacy for microorganisms with an MIC of ≥1 mg/L.

Conclusion: A population PK model of gentamicin, linked to a utility function, estimated an optimal initial dose of 8.3 mg/kg, which provides an average probability of efficacy of 85%, 15% of the maximal possible gentamicin accumulation and acceptable bacterial killing.


1.             Moore RD et al. (1987). J Infect Dis 155:93-99.,

2.             Kashuba AD et al. (1999). Antimicrob Agents Chemother 43:623-629.

3.             Giuliano RA et al. (1986). J Pharmacol Exp Ther 236:470-475.

4.             Mohamed AF et al. (2012). Antimicrob Agents Chemother 56:179-188.

5.             Zhuang L et al. (2016). J Antimicrob Chemother 71:1012-1021.

6.             Rhodin M et al. (2009). Pediatr Nephrol 24:67-76