Abstract
1631
Objectives Administering unlabelled antibody has improved radioimmunotherapy using anti-CD45 antibody. Accurate predictions of the therapeutical biodistribution depending on different preloads for each patient can help to optimize RIT. In this work, the influence of mono and bivalent binding on the predicted anti-CD45 antibody biodistribution is investigated.
Methods Two physiologically based pharmacokinetic models were developed. Model A uses effective monovalent values for the association and dissociation of antibody and antigen. Model B explicitly allows mono- and bivalent binding. The models were fitted to In-111 labeled anti-CD45 antibody time activity data (5 AML patients) of red marrow, liver, spleen, total body and blood (0 h-144 h). The corrected Akaike information criterion (AICc) was employed for model selection. Furthermore, the amount of antibody leading to the most favorable biodistribution was determined.
Results All fits were excellent. Model B was selected best according to the AICc. Most interestingly, for model A (23±8) mg and B (25±9) mg approximately the same optimal preload for all patients was identified. However, the estimated parameters and the extent of the biodistribution improvement differ. The mean values and standard deviations of the residence times for red marrow and liver are (37.7±8.0; 7.8±2.5) h and (32.5±10.8; 5.0±2.7) h for model A and B, respectively.
Conclusions Despite the very important fact that for model A and B the amount of antibody leading to the most favorable biodistribution was approximately the same, the estimated parameters and the predicted biodistribution differed. To determine which model is best in predicting the biokinetics of the therapy further measurement are required.
Research Support German research foundation (DFG GL 236/7-1).
- © 2009 by Society of Nuclear Medicine