An adaptive biosystems engineering approach towards modeling the soluble-to-insoluble phase transition of clofazimine

Clofazimine (CFZ) is a weakly basic, small-molecule antibiotic used for the treatment of mycobacterial infections including leprosy and multidrug-resistant tuberculosis. Upon prolonged oral administration, CFZ precipitates and accumulates within macrophages throughout the host. To model the pharmacokinetics of CFZ, the volume of distribution (Vd) was considered as a varying parameter that increases with continuous drug loading. Fitting the time-dependent change in drug mass and concentration data obtained from CFZ-treated mice, we performed a quantitative analysis of the systemic disposition of the drug over a 20-week treatment period. The pharmacokinetics data were fitted using various classical compartmental models sampling serum and spleen concentration data into separate matrices. The models were constructed in NONMEM together with linear and nonlinear sigmoidal expansion functions to the spleen compartment to capture the phase transition in Vd. The different modeling approaches were compared by Akaike information criteria, observed and predicted concentration correlations, and graphically. Using the composite analysis of the modeling predictions, adaptive fractional CFZ sequestration, Vd and half-life were evaluated. When compared to standard compartmental models, an adaptive Vd model yielded a more accurate data fit of the drug concentrations in both the serum and spleen. Including a nonlinear sigmoidal equation into compartmental models captures the phase transition of drugs such as CFZ, greatly improving the prediction of population pharmacokinetics and yielding further insight into the mechanisms of drug disposition.