Mathematical modeling and systems pharmacology of tuberculosis: Isoniazid as a case study.

Tuberculosis (TB) treatment needs to be optimized as it is currently long and associated with increasing drug resistance. The antimycobacterial effect of isoniazid (INH) is characterized by a biphasic kill curve, whose causes are still debated. In this work, we developed a complete mathematical model describing the time-course of TB infection and its treatment by INH in human lung. This model was based on a pharmacokinetic model, a pharmacodynamic model and a pathophysiological model. It was used to simulate the antibacterial effect of INH during the first days of therapy. This full model adequately reproduced some qualitative and quantitative properties of the early bactericidal activity of INH observed in TB patients. The kill curves simulated with the model reproduced the biphasic killing effect of INH and the predicted declines in extracellular bacteria were comparable to clinical data. A sensitivity analysis provided interesting insights regarding the biphasic kill curve. The first phase appeared to be essentially driven by the drug effect. In the second phase, while drug pharmacology was the major determinant of the antibacterial effect, a slight influence of the dynamics of infected macrophages was also observed. This work permits to formulate hypotheses for optimizing the efficacy of TB drug candidates and confirms the utility of mathematical modeling to generate new assumptions for TB research.