Regimen-Specific Population Pharmacokinetics of Isoniazid with and Without Rifamycin: A Bayesian Modeling Analysis of...

Background: Isoniazid (INH) remains a cornerstone of tuberculosis (TB) prevention and treatment, administered either as monotherapy or in combination with rifamycin-containing regimens. In this study, the preventive regimens analyzed included 6 months of daily INH monotherapy (6H), 3 months of daily INH plus rifampicin (3HR), and 3 months of twice-weekly INH plus rifapentine (3H2P2). Despite the adoption of shorter-course regimens, substantial inter-individual variability (IIV) in INH exposure persists, potentially impacting both therapeutic efficacy and toxicity. A quantitative, regimen-specific characterization of INH pharmacokinetics is therefore critical to support model-informed dosing strategies. Methods: Population pharmacokinetic models were developed separately for INH administered as 6H, 3HR, and 3H2P2. The models characterized absorption, clearance, and IIV of INH, while accounting for co-administered rifamycin. The effects of N-acetyltransferase 2 (NAT2) acetylator phenotype and relevant clinical covariates were systematically evaluated. Model performance was assessed using goodness-of-fit diagnostics, posterior predictive checks, and visual predictive checks. Population pharmacokinetic models were developed using a Bayesian nonlinear mixed-effects framework implemented in Stan through CmdStanR version 0.9.0, with additional data processing, statistical summaries, and visualization performed using R version 4.2.3. Results: INH pharmacokinetics were adequately described by regimen-specific models, revealing distinct differences in absorption and variability across regimens. Typical INH apparent oral clearance (CL/F) estimates were 21.83 L/h for 6H, 26.32 L/h for 3HR, and 25.98 L/h for 3H2P2. NAT2 phenotype was a major determinant of INH clearance across regimens: compared with intermediate acetylators, slow acetylators showed 35.4% lower CL/F, whereas fast acetylators showed 48.4% higher CL/F, indicating higher INH exposure in slow acetylators and lower exposure in fast acetylators. Body weight also influenced INH pharmacokinetics. Co-administration with rifampicin or rifapentine influenced INH pharmacokinetics in a manner consistent with reduced exposure in rifamycin-containing regimens, particularly among fast acetylators. Conclusions: Regimen-specific population pharmacokinetic modeling elucidated clinically relevant differences in INH exposure across commonly used preventive and treatment regimens. These findings highlight the importance of accounting for regimen- and genotype-specific effects when optimizing INH dosing and provide a quantitative framework for future model-informed precision dosing approaches in TB care.