Discussion
In the present study, AUC prediction based on C6 was the most accurate one-point blood sampling strategy. Watanabe et al .9 reported that the AUC of busulfan was best predicted by C6 in pediatric hematopoietic stem cell transplant recipients, supporting the results of the present study. The reason for the poor predictive performance of AUC based on C1 could be because the individual differences in the absorption phase were smaller than in the elimination phase due to the intravenous administration of busulfan. C2 represents the blood sampling immediately after the end of busulfan infusion and may reflect the sampling of the distribution phase. Therefore, the blood concentrations may significantly change within a few minutes, which might explain the weaker relationship between the predicted AUC based on C2 and the actual AUC.
In the present study, the predictive performance of AUC based on C6 was lower (r2 = 0.789) than that reported by Watanabe et al. 9(r2 = 0.929). More than 60% of all patients weighed ≥23 kg in that study, whereas in our study, 24% of the patients weighed ≥23 kg, which might have reduced the accuracy of predicted AUC based on C6. In fact, from the predictive performance of AUC performance by body weight, the predicted AUC based on C6 was highly precise in adolescents.
In infants and young children, the accuracy of AUC predicted by one-point sampling with C6 was poor. Kishimoto et al .10 reported that errors in AUC prediction based on C6 were higher in young children and infants using the Monte Carlo simulation in the population pharmacokinetic model. Furthermore, McCune et al .6 reported that the accuracy of prediction for busulfan to reach the therapeutic range was lower in patients younger than five years than in adolescent patients using the population pharmacokinetic model. Therefore, in early childhood, the busulfan pharmacokinetics significantly varies among individuals, and predicting AUC by the one-point sampling strategy may not be sufficient. In the present study, the AUC prediction based on C3 and C6 using the two-point sampling strategy was acceptable even for infants and young children. In the validation cohort, more than half of the patients had more than 15% errors with the one-point sampling strategy compared to the actual AUC. Errors >15% are a deviation from the guidelines on bioanalytical method validation in pharmaceutical development,11 which states that the accuracy and precision should be within ±15% deviation of the theoretical concentration. Thus, in infants and young children, the AUC should be predicted using the two-point sampling strategy.
In the one-point sampling strategy, the LSS equation for AUC was 2.866 C6 + 108.9 in adolescent patients. Consequently, the target C6 was 360–500 ng/mL for myeloablative conditioning (target AUC, 75–100 mg/L × h) and 200–300 ng/mL for reduced-intensity conditioning (target AUC, 45–65 mg/L × h). In contrast, in the two-point sampling strategy, the target C3 depended on C6 and target AUC. It should be noted that the accuracy of the predicted AUC by LSS will decrease if the blood sampling time is significantly different from the correct sampling time. Thus, it is important to pay close attention to blood sampling times.
Concomitant medications that may affect the busulfan pharmacokinetics should be considered in the interpretation of our findings. Kangarlooet al .12 reported that phenytoin, an anticonvulsant agent, increased the busulfan clearance. In the present study, levetiracetam or clonazepam were used as anticonvulsants, which do not affect the busulfan pharmacokinetics.13,14Therefore, the anticonvulsant agent did not have a significant effect on the busulfan pharmacokinetics in the present study.
The present study also has certain limitations. It remains unclear why busulfan pharmacokinetics varies significantly in infants and young children. Polymorphisms in glutathione S -transferase (GST )15 affect busulfan pharmacokinetics. Additionally, age-dependent variations in intrinsic busulfan clearance were reported to be associated with higher GST activity16; thus, differences in the GST expression might explain individual differences in busulfan pharmacokinetics. Further investigation is warranted to establish the parameters for personalized administration of busulfan.
In conclusion, the AUC of busulfan could be predicted based on C6 in adolescent patients. However, for infants and young children, there was substantial inter-individual variation in busulfan pharmacokinetics and the two-point sampling strategy was necessary for accurate AUC prediction.