Abstract
L-4-[((10)B)]Boronophenylalanine (BPA) is an amino acid analogue with a boron-10 moiety. It is most widely used as a boron carrier in boron neutron capture therapy. In this study, a Bayesian predictive platform of blood boron concentration based on a BPA pharmacokinetic (PK) model was developed. This platform is user-friendly and can predict the individual boron PK and optimal time window for boron neutron capture therapy in a simple way. The present study aimed to establish a PK model of L-4-boronophenylalanine and develop a Bayesian predictive platform for blood boron PKs for user-friendly estimation of boron concentration during neutron irradiation of neutron capture therapy. Whole blood boron concentrations from seven previous reports were graphically extracted and analyzed using the nonlinear mixed-effects modeling (NONMEM) approach. Model robustness was assessed using nonparametric bootstrap and visual predictive check approaches. The visual predictive check indicated that the final PK model is able to adequately predict observed concentrations. The Shiny package was used to input real-time blood boron concentration data, and during the following irradiation session blood boron was estimated with an acceptably short calculation time for the determination of irradiation time. Finally, a user-friendly Bayesian estimation platform for BPA PKs was developed to optimize individualized therapy for patients undergoing BNCT.