Abstract
Busulfan (BU) is widely used in conditioning regimens before allogeneic or autologous bone marrow transplantation. BU pharmacokinetic (PK) studies have suggested that individualization of BU doses based on the area under the concentration curve (AUC) is necessary for optimal BU systemic exposure. Since standard PK measurements need multiple blood samples, various limited sampling methods (LSM) for the estimation of AUC have been proposed to reduce the number of blood samples. However, it is not clear under which conditions the estimations succeed or fail. Also, none of the existing LSMs for BU estimate the variance of AUC estimator. To solve the first problem, we introduce a systematic validation method by use of a virtual validation set generated based on the identified PK model and the distribution of its parameters. To introduce a more accurate LSM, we propose a new method for AUC estimation, in which a curve that best approximates the measured data is searched from the set of pre-generated model cases. We evaluated this estimation method and found that it has virtually no bias. For the latter variance problem, we introduce a method for calculating the distribution of an AUC estimator based on the measured distribution of within-patient errors and by the use of Monte Carlo simulation, from which we develope bootstrap 95% percentiles. This new method for the AUC estimation can also be applied to other dose-finding studies and their applications.
