Towards Model-Informed Precision Dosing of High-Dose Rifampicin: A Parsimonious Population Pharmacokinetic Model Capturing Saturation and Autoinduction in Tanzanian and South African Patients with Tuberculosis

Rifampicin, a cornerstone of tuberculosis treatment, exhibits complex pharmacokinetics with variable absorption, saturable elimination and time-dependent autoinduction, together with marked inter- and intra-individual variability in drug exposure. These properties challenge standard weight-based dosing and underscore the need for individualized strategies, particularly in high-burden settings like Tanzania. While therapeutic drug monitoring (TDM) supported by model-informed precision dosing (MIPD) is being implemented, few of the existing population pharmacokinetic (popPK) models includes Tanzanians and are parsimonious. This study aimed to develop a parsimonious popPK model of rifampicin that successfully captures both its autoinductive and saturable elimination components, achieving predictive performance comparable to more complex reference models described in the literature. Rifampicin concentrations data from 42 Tanzanian and 92 South African patients, covering doses up to 50 mg/kg over 2 weeks were used to develop the model, which was further benchmarked against literature reference models. The chosen model employed a one-compartment structure with Michaelis-Menten elimination and a time-dependant E max -based autoinduction function. The final model demonstrated individual Bayesian predictive performance comparable to two complex reference popPK models, and confirmed substantial inter-individual variability in exposure, with exposures differing by up to fivefold within the same weight-based dosing cohort. These findings support the implementation of MIPD and TDM to improve efficacy and safety in TB treatment rather than relying on a uniform dosing approach based on bodyweight. This model provides a robust foundation for individualized dosing and the deployment of clinical decision support tools in high TB burden settings.