Population pharmacokinetics, optimised design and sample size determination for rifampicin, isoniazid, ethambutol and...

The current first-line therapy for drug-susceptible tuberculosis consists of rifampicin (RIF), isoniazid (INH), pyrazinamide (PZA) and ethambutol (EMB). In this study, we determined the population pharmacokinetics (PopPK) of RIF, INH, EMB and PZA using original experimental sampling designs for single-dose intravenous (IV) and single- and multiple-dose oral administration studies in the mouse model, and used these PopPK models to develop and evaluate new, more informative sampling designs with the aim of reducing the number of animals required for each drug. The RIF, INH, EMB and PZA blood concentrations after single oral and IV doses and multiple-dose oral administrations based on the original designs were used in the PopPK analysis using NONMEM software. The final PopPK models described the data well. Stochastic simulation and estimation were used to optimise the designs. The relative bias and relative imprecision of each pharmacokinetic parameter for each drug were derived and assessed to choose the final designs. The final single-dose IV and oral designs included up to eight samples per mouse with a total of 24 mice required for RIF and EMB and 33 mice for INH and PZA. In the new multiple-dose (zipper) oral designs, the mice were divided into two groups of three per dose, and four samples were taken from each mouse to cover all seven or eight sampling time points. The final number of mice required for the multiple-dose oral designs was 30 for RIF, INH and EMB, 36 for PZA. The number of mice required in the new designs for RIF, INH and EMB was decreased by up to 7-fold and the relative bias and relative imprecision in the parameter estimates were at least similar to those in the original designs.