Population pharmacokinetics of doxorubicin: establishment of a NONMEM model for adults and children older than 3 years

The aim of the current investigation was to develop a population pharmacokinetic model for doxorubicin and doxorubicinol that could provide improved estimated values for the pharmacokinetic parameters clearance of doxorubicin, volume of distribution of the central compartment, clearance of doxorubicinol and volume of distribution of the metabolite compartment for adults and children older than 3 years. A further aim was to investigate the potential influence of the covariates body surface area, body weight, body height, age, body mass index, sex and lean body mass on the pharmacokinetic parameters.

Three different datasets, two containing data from adults and one containing data from adults and children, were merged and the combined dataset was analysed retrospectively. In total, the combined dataset contained 934 doxorubicin and 935 doxorubicinol plasma concentrations from 82 patients [64 adults and 18 children (<18 years)]. With this combined dataset, a population pharmacokinetic model was developed, using NONMEM^® 7.2 and a predefined model-building strategy. Different structural models, error models and estimation methods were tested, and the inter-individual and the inter-occasion variability (variability between separate (two or three) doxorubicin infusions) were tested. Using a subset of 52 patients, the influence of different covariates on the pharmacokinetic parameters was investigated. The pharmacokinetic parameter estimates obtained from doxorubicin concentrations with the best model were fixed, and an additional compartment for doxorubicinol was added to the model. With the final model for both substances, a potential age dependency and body mass index dependency of the clearance of doxorubicin and doxorubicinol as well as of the volumes of distribution of the central and the metabolite compartment were evaluated.

A four-compartment model best described the doxorubicin and doxorubicinol data of the combined dataset. This model included a proportional residual error model and an inter-individual variability on the clearance of doxorubicin, on the inter-compartmental clearances of the peripheral compartments, on the clearance of doxorubicinol and on the volumes of distribution of the central, one peripheral and the metabolite compartment. Furthermore, the body surface area as covariate on all pharmacokinetic parameters and an inter-occasion variability for the clearance of doxorubicin and the volume of distribution of the central compartment were incorporated in the model. For a patient with the body surface area of 1.8 m², the clearance of doxorubicin was 53.3 L/h (inter-individual variability 31 %, inter-occasion variability 13 %) and the volume of distribution of the central compartment was 17.7 L (inter-individual variability 19 %, inter-occasion variability 21 %), respectively. The residual variability of the model was 22 % for doxorubicin and 26 % for doxorubicinol. The clearance of doxorubicinol was estimated at 44 L/h (inter-individual variability 50 %) and the volume of distribution of the metabolite compartment at 1,150 L (inter-individual variability 57 %). The evaluation of a possible age dependency and body mass index dependency showed a trend to a smaller volume of distribution of the central compartment (normalised to the body surface area) and a higher volume of distribution of the metabolite compartment (normalised to the body weight) in younger patients.

A four-compartment NONMEM^® model for doxorubicin and doxorubicinol adequately described the plasma concentrations in adults and children (>3 years). No pronounced effects of age on the clearance of doxorubicin or doxorubicinol were found, and the analysis did not support the modification of the dosing strategies presently used in children and adults.