Evaluation of LSSs
First, the regression equation was calculated from individual busulfan
concentrations at each sampling point and the actual
AUC0–∞ obtained from the pharmacokinetic modeling
software. Second, the predicted AUC0–∞ was calculated
from the regression equation and individual busulfan concentrations.
Finally, the relationship between the predicted AUC0–∞and actual AUC0–∞ was evaluated using Pearson’s
correlation coefficient (r2 ). LSSs using two or
three busulfan sampling points were developed by multiple regression
analysis. The precision of the predictive performance of busulfan AUCs
was assessed by calculating the mean absolute percentage error (MAPE).
Additionally, subgroup analysis was conducted based on body weight
(<9, 9–16, 16–23, 23–34, and >34 kg). To
verify the utility of LSSs in the present study, the predicted
AUC0–∞ was calculated using the developed LSS and its
predictive performance was evaluated in seven patients in a cohort
separate from the first cohort that was used to develop the LSS. All
statistical analyses were performed using SPSS Statistics ver.23 (SPSS
IBM Japan, Tokyo, Japan).