Evaluation of Existing Limited Sampling Models for Therapeutic Drug Monitoring of Busulphan in Children and Adults Undergoing SCT.

Busulphan (Bu) is used in high dose conditioning regimen prior to stem cell transplantation. Bu has a narrow therapeutic window and over- and under-dosing may have a fatal outcome. Bu pharmacokinetics and pharmacodynamics were extensively studied and wide inter- and intra-individual variation was found. Several limited sampling models (LSM) have been developed for Bu administered orally to simplify therapeutic drug monitoring and consequently dose adjustment. The aim of this study was to evaluate the existing LSM in adults and children undergoing conditioning regimen before SCT. Seventy-four patients (62 adults and 12 children) with malignant and non-malignant diseases were analysed. Plasma was sampled at 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 5 and 6 hours after the first dose of Bu. Bu was determined using gas chromatography with electron capture detection. The area under the plasma concentration-time curve (AUC) for time interval 0 to 6 hours was determined using Winnonlin program and trapezoidal rule. Results were compared to the estimated AUCs using LSMs (Vassal 1992, Schuler 1994, Hassan 1996, Chatergoon 1997). The best correlation between the AUCs determined using trapezoidal rule and 3-points model by Schuler was found (R²=0.95 for all patients, R²=0.97 for children and R²=0.94 for adults). In children, a correlation between AUCs determined with trapezoidal rule and following LSMs was found: 2-points LSM by Schuler (R²=0.94), LSM by Hassan (R²=0.94), LSM by Vassal (R²=0.81) and 3 of 5 LSMs by Chatergoon (R²=0.85, 0.88 and 0.87, resp.). AUCs in children determined using Winnonlin showed good correlation with both Schuler’s models, model by Hassan and one of 4-points models by Chatergoon. However, the correlation between the AUCs determined using trapezoidal rule and Winnonlin was good in children (R²=0.98), but not in adults (R²=0.65). Thus, several limited sampling models are suitable for AUC estimation in children, while there is only one suitable model for adults. This conclusion is made with reservation that even the trapezoidal rule may underestimate the real AUC dependent on sampling density.