Modeling Treatment Response in Tuberculosis Early Bactericidal Activity Trials

Area of Science:

• Pharmacology and Microbiology
• Clinical Trial Biomarker Development

Background:

• Culture-based bacterial load monitoring in tuberculosis (TB) clinical trials is time-consuming and prone to data gaps.
• Non-culture-based assays, such as the Tuberculosis Molecular Load Bacterial Assay (TB-MBLA), offer a potential solution for more efficient TB drug development.

Purpose of the Study:

• To evaluate TB-MBLA as a biomarker in early bactericidal activity (EBA) studies for TB.
• To explore the utility of combining TB-MBLA with other biomarkers using joint modeling for deeper biological insights.

Main Methods:

• TB-MBLA data were generated from sputum samples of 78 patients in the PanACEA BTZ-043 Phase Ib/IIa trial.
• Probabilistic principal component analysis (pPCA) was used to derive a summary measure (PC1) from TB-MBLA, colony forming units (CFU), and time-to-positivity (TTP) data.
• Linear and non-linear mixed models were applied to re-evaluate dose-response and pharmacokinetic-pharmacodynamic (PK-PD) analyses using TB-MBLA and PC1 values.

Main Results:

• TB-MBLA did not show a detectable exposure-response effect in PK-PD analysis, unlike CFU and TTP.
• Combining biomarkers revealed a significant, albeit less pronounced, Emax exposure-response between days 0-3 compared to using CFU or TTP alone.
• pPCA successfully integrated CFU and TTP into a joint latent component, enhancing treatment effect detection compared to individual biomarkers.

Conclusions:

• While TB-MBLA alone did not show a direct PK-PD exposure-response, its combination with other biomarkers improved EBA assessment.
• Joint modeling using pPCA provides a valuable framework for integrating multiple bacterial load markers, enhancing the detection of treatment efficacy in TB trials.
• This study provides novel EBA data for BTZ-043, highlighting the value of emerging bacteriological markers in TB clinical trial design.