Physiologically based pharmacokinetic study on suramin

3991 Purpose: Suramin failed in multiple clinical trials as a chemotherapeutic agent. However, it was proved to be a chemosensitizer at low or non-toxic concentrations (10 to 50 μM) (Villalona-Calero MA, et al., Clin. Cancer Res., 2003, Aug 15; 9(9): 3303-11), which are 4 to 20 fold lower than the normal clinically desired plasma concentration (200μM). Present study aims to understand the suramin distribution characteristics in this low or non-toxic concentration range at its action site: tissues. Method: Suramin distribution in blood was done with ultracentrifuge method. And a series of tissue disposition kinetic study was performed in nude mice after 10, 30, or 200 mg/kg i.v. bolus dose. Plasma and tissue concentrations were measured with HPLC. Results: Extensive tissue binding and extremely long tissue retention of suramin were found in most tissues. Although plasma pharmacokinetics showed a dose linear pattern, suramin distribution in tissues is very likely dose nonlinear in many organs (kidney, skin, muscle, liver, and gut), where higher dose generates a lower normalized AUC value (AUC/dose). Accordingly, apparent volume of distribution decreases at higher dose (2165.8 ml at 10 mg/kg dose, 1754.3 ml at 30 mg/kg dose, and 545.0 ml at 200 mg/kg) and the saturable binding is possibly the cause. A whole-body physiologically based pharmacokinetic (PBPK) model was therefore established to better understand this process. The tissue distribution is best described with either a Michaelis-Menten kinetics based tissue binding model (heart, muscle, kidney, gut, spleen, and skin; model 1), a linear model with deep tissue binding sub-compartment (liver; model 2), or a Michaelis-Menten kinetics based tissue binding model with a supplemental tissue binding sub-compartment (lung; model 3). Simulation results agree well with experimental results in plasma and all organs. Extremely high drug binding capacity was found in kidney (BM = 1482 μg-eq/ml), spleen (BM = 1213 μg-eq/ml), and gut (BM = 3317 μg-eq/ml). The model was then up-scaled to human beings by modifying physiological parameters, protein binding value and elimination clearance. The simulated plasma concentration profiles correlate closely to experimentally measured values from our phase I clinical trial, further suggesting the validity of current model. Conclusions: Overall, the PBPK model provides a powerful tool to quantitatively link preclinical and clinical data.