Background and Objectives
Even though lithium has been used for the treatment of bipolar disorder for several decades, its toxicities are still being reported. The major limitation in the use of lithium is its narrow therapeutic window. Several methods have been proposed to predict lithium doses essential to attain therapeutic levels. One of the methods used to guide lithium therapy is population pharmacokinetic approach which accounts for inter- and intra-individual variability in predicting lithium doses. Several population pharmacokinetic studies of lithium have been conducted. The objective of this review is to provide information on population pharmacokinetics of lithium focusing on nonlinear mixed effect modeling approach and to summarize significant factors affecting lithium pharmacokinetics.
A literature search was conducted from PubMed database from inception to December, 2016. Studies conducted in humans, using lithium as a study drug, providing population pharmacokinetic analyses of lithium by means of nonlinear mixed effect modeling, were included in this review.
Twenty-four articles were identified from the database. Seventeen articles were excluded based on the inclusion and exclusion criteria. A total of seven articles were included in this review. Of these, only one study reported a combined population pharmacokinetic–pharmacodynamic model of lithium. Lithium pharmacokinetics were explained using both one- and two-compartment models. The significant predictors of lithium clearance identified in most studies were renal function and body size. One study reported a significant effect of age on lithium clearance. The typical values of lithium clearance ranged from 0.41 to 9.39 L/h. The magnitude of inter-individual variability on lithium clearance ranged from 12.7 to 25.1%. Only two studies evaluated the models using external data sets.
Model methodologies in each study are summarized and discussed in this review. For future perspective, a population pharmacokinetic–pharmacodynamic study of lithium is recommended. Moreover, external validation of previously published models should be performed.