PT-013 - USING MODEL-BASED SIMULATIONS TO ALIGN MOUSE ANTIBIOTIC EXPOSURE WITH HUMAN CLINICAL EFFICACY FOR EXPLORATION OF COMBINATION THERAPY IN TUBERCULOSIS

A. Singh¹, I. Alshiraihi², S. Maloney³, A. Hickey³, G. Mercedes⁴, B. Meibohm⁵; ¹The University of Tennesse Heath Sciences, The University of Tennessee Health Sciences, USA, ²Colorado State University, Fort Collins, USA, ³RTI International, North Carolina, USA, ⁴Fort Collins, Colorado, USA, ⁵The University of Tennessee Health Sciences, TN, USA.

Background: Preclinical experiments frequently utilize well-established drugs to evaluate various combination therapies and assess early efficacy, for example in tuberculosis therapy. Historically, the doses administered to animals in these studies are often selected based on literature values or arbitrary estimates, which may not accurately reflect the drug’s clinical efficacy. To address this issue, we employed a modeling and simulation-based approach to precisely determine appropriate dosing regimens for pre-clinical species, leveraging PK data from existing literature and advanced simulation tools.

Methods: We began by extracting detailed mouse PK profiles for bedaquiline from the literature, using bootstrap techniques to include interindividual variability. Using the Monolix suite, we created nonlinear mixed-effects models that captured the drug’s behavior in mice. Simulx was then employed to conduct simulations across various dosing regimens (ranging from 10 mg/kg to 50 mg/kg) to explore their impact on drug exposure (AUC). Our simulations aimed to determine a dosing regimen for mice that would replicate the PK profile observed in humans, targeting an area under the curve for free drug (fAUC), which is a well-known efficacy index of bedaquiline in tuberculosis therapy.

Results: The optimal mice dose of 25-30 mg/kg oral once daily was selected based on its ability to achieve an AUC/MIC value that aligns with human efficacious AUC/MIC value, ensuring that preclinical dosing would provide clinically relevant exposure. The plasma protein binding between mice and humans was found to be the same which helps in simplifying the translation process. However, it is important to note that our model did not account for the formation of M2 and M3 metabolites in mice, contributing approximately 30% to the efficacy observed in humans. This omission represents a potential area for future refinement.

Conclusion: In conclusion, our study highlights a model-based approach to translate human clinical doses to preclinical mouse doses through exposure matching. Thus, leveraging detailed PK data and sophisticated simulation techniques, enhances the relevance of preclinical studies, facilitating more effective evaluation of combination therapies and potentially improving the translation of research findings from mice to humans.