Objectives: Bedaquiline is the first drug of a new class approved for the treatment of TB in decades. Bedaquiline is metabolized by cytochrome P450 (CYP) 3A4 to a less-active M2 metabolite. Its terminal half-life is extremely long (5–6 months), complicating evaluations of drug–drug interactions. Rifampicin and rifapentine, two anti-TB drugs now being optimized to shorten TB treatment duration, are potent inducers of CYP3A4. This analysis aimed to predict the effect of repeated doses of rifampicin or rifapentine on the steady-state pharmacokinetics of bedaquiline and its M2 metabolite from single-dose data using a model-based approach.Methods: Pharmacokinetic data for bedaquiline and M2 were obtained from a Phase I study involving 32 individuals each receiving two doses of bedaquiline, alone or together with multiple-dose rifampicin or rifapentine. Sampling was performed over 14 days following each bedaquiline dose. Pharmacokinetic analyses were performed using non-linear mixed-effects modelling. Models were used to simulate potential dose adjustments.Results: Rifamycin co-administration increased bedaquiline clearance substantially: 4.78-fold [relative standard error (RSE) 9.10%] with rifampicin and 3.96-fold (RSE 5.00%) with rifapentine. Induction of M2 clearance was equally strong. Average steady-state concentrations of bedaquiline and M2 are predicted to decrease by 79% and 75% when given with rifampicin or rifapentine, respectively. Simulations indicated that increasing the bedaquiline dosage to mitigate the interaction would yield elevated M2 concentrations during the first treatment weeks.Conclusions: Rifamycin antibiotics reduce bedaquiline concentrations substantially. In line with current treatment guidelines for drug-susceptible TB, concomitant use is not recommended, even with dose adjustment.

Tuberculosis remains a huge public health problem and the prolonged treatment duration obstructs effective tuberculosis control. Higher rifampicin doses have been associated with better bactericidal activity, but optimal dosing is uncertain. This analysis aimed to characterize the relationship between rifampicin plasma exposure and treatment response over 6 months in a recent study investigating the potential for treatment shortening with high-dose rifampicin. Data were analyzed from 336 patients with pulmonary tuberculosis (97 with pharmacokinetic data) treated with rifampicin doses of 10, 20, or 35 mg/kg. The response measure was time to stable sputum culture conversion (TSCC). We derived individual exposure metrics with a previously developed population pharmacokinetic model of rifampicin. TSCC was modeled using a parametric time-to-event approach, and a sequential exposure-response analysis was performed. Higher rifampicin exposures increased the probability of early culture conversion. No maximal limit of the effect was detected within the observed range. The expected proportion of patients with stable culture conversion on liquid medium at week 8 was predicted to increase from 39% (95% confidence interval, 37%–41%) to 55% (49%–61%), with the rifampicin area under the curve increasing from 20 to 175 mg/L·h (representative for 10 and 35 mg/kg, respectively). Other predictors of TSCC were baseline bacterial load, proportion of culture results unavailable, and substitution of ethambutol for either moxifloxacin or SQ109. Increasing rifampicin exposure shortened TSCC, and the effect did not plateau, indicating that doses >35 mg/kg could be yet more effective. Optimizing rifampicin dosage while preventing toxicity is a clinical priority.

Pharmacometrics (PM) and machine learning (ML) are both valuable for drug development to characterize pharmacokinetics (PK) and pharmacodynamics (PD). Pharmacokinetic/pharmacodynamic (PKPD) analysis using PM provides mechanistic insight into biological processes but is time- and labor-intensive. In contrast, ML models are much quicker trained, but offer less mechanistic insights. The opportunity of using ML predictions of drug PK as input for a PKPD model could strongly accelerate analysis efforts. Here exemplified by rifampicin, a widely used antibiotic, we explore the ability of different ML algorithms to predict drug PK. Based on simulated data, we trained linear regressions (LASSO), Gradient Boosting Machines, XGBoost and Random Forest to predict the plasma concentration-time series and rifampicin area under the concentration-versus-time curve from 0–24 h (AUC_0–24h) after repeated dosing. XGBoost performed best for prediction of the entire PK series (R²: 0.84, root mean square error (RMSE): 6.9 mg/L, mean absolute error (MAE): 4.0 mg/L) for the scenario with the largest data size. For AUC_0–24h prediction, LASSO showed the highest performance (R²: 0.97, RMSE: 29.1 h·mg/L, MAE: 18.8 h·mg/L). Increasing the number of plasma concentrations per patient (0, 2 or 6 concentrations per occasion) improved model performance. For example, for AUC_0–24h prediction using LASSO, the R² was 0.41, 0.69 and 0.97 when using predictors only (no plasma concentrations), 2 or 6 plasma concentrations per occasion as input, respectively. Run times for the ML models ranged from 1.0 s to 8 min, while the run time for the PM model was more than 3 h. Furthermore, building a PM model is more time- and labor-intensive compared with ML. ML predictions of drug PK could thus be used as input into a PKPD model, enabling time-efficient analysis.

Albumin concentration and body weight are altered in patients with multidrug‐resistant tuberculosis (MDR‐TB) and change during the long treatment period, potentially affecting drug disposition. We here describe the pharmacokinetics (PKs) of the novel anti‐TB drug bedaquiline and its metabolite M2 in 335 patients with MDR‐TB receiving 24 weeks of bedaquiline on top of a longer individualized background regimen. Semiphysiological models were developed to characterize the changes in weight and albumin over time. Bedaquiline and M2 disposition were well described by three and one‐compartment models, respectively. Weight and albumin were correlated, typically increasing after the start of treatment, and significantly affected bedaquiline and M2 plasma disposition. Additionally, age and race were significant covariates, whereas concomitant human immunodeficiency virus (HIV) infection, sex, or having extensively drug‐resistant TB was not. This is the first population model simultaneously characterizing bedaquiline and M2 PKs in its intended use population. The developed model will be used for efficacy and safety exposure‐response analyses.

Background: Intensified antimicrobial treatment with higher rifampicin doses may improve outcome of tuberculous meningitis, but the desirable exposure and necessary dose are unknown. Our objective was to characterize the relationship between rifampicin exposures and mortality in order to identify optimal dosing for tuberculous meningitis.Methods: An individual patient meta-analysis was performed on data from 3 Indonesian randomized controlled phase 2 trials comparing oral rifampicin 450 mg (~10 mg/kg) to intensified regimens including 750–1350 mg orally, or a 600-mg intravenous infusion. Pharmacokinetic data from plasma and cerebrospinal fluid (CSF) were analyzed with nonlinear mixed-effects modeling. Six-month survival was described with parametric time-to-event models.Results: Pharmacokinetic analyses included 133 individuals (1150 concentration measurements, 170 from CSF). The final model featured 2 disposition compartments, saturable clearance, and autoinduction. Rifampicin CSF concentrations were described by a partition coefficient (5.5%; 95% confidence interval [CI], 4.5%–6.4%) and half-life for distribution plasma to CSF (2.1 hours; 95% CI, 1.3–2.9 hours). Higher CSF protein concentration increased the partition coefficient. Survival of 148 individuals (58 died, 15 dropouts) was well described by an exponentially declining hazard, with lower age, higher baseline Glasgow Coma Scale score, and higher individual rifampicin plasma exposure reducing the hazard. Simulations predicted an increase in 6-month survival from approximately 50% to approximately 70% upon increasing the oral rifampicin dose from 10 to 30 mg/kg, and predicted that even higher doses would further improve survival.Conclusions: Higher rifampicin exposure substantially decreased the risk of death, and the maximal effect was not reached within the studied range. We suggest a rifampicin dose of at least 30 mg/kg to be investigated in phase 3 clinical trials.

Bedaquiline has been shown to reduce time to sputum culture conversion (SCC) and increase cure rates in patients with drug-resistant TB, but the influence of drug exposure remains uncharacterized. To investigate whether an exposure–response relationship could be characterized by making better use of the existing information on pharmacokinetics and longitudinal measurements of mycobacterial load. Quantitative culture data in the form of time to positivity (TTP) in mycobacterial growth indicator tubes obtained from a randomized placebo-controlled Phase IIb registration trial were examined using non-linear mixed-effects methodology. The link to individual bedaquiline exposures and other patient characteristics was evaluated. The developed model included three simultaneously fitted components: a longitudinal representation of mycobacterial load in patients, a probabilistic component for bacterial presence in sputum samples, and a time-to-event model for TTP. Data were described adequately, and time to SCC was well predicted. Individual bedaquiline exposure was found to significantly affect the decline in mycobacterial load. Consequently, the proportion of patients without SCC at week 20 is expected to decrease from 25% (95% CI 20%–31%) without bedaquiline to 17% (95% CI 13%–21%), 12% (95% CI 8%–16%) and 7% (95% CI 4%–11%), respectively, with half the median, median and double the median bedaquiline exposure observed in patients with standard dosing. Baseline bacterial load and level of drug resistance were other important predictors. To our knowledge, this is the first successful description of bedaquiline’s exposure–response relationship and may be used when considering dose optimization. Characterization of this relationship was possible by integrating quantitative information in existing clinical data using novel models.