Skip to main content
NCBI home page
Nature Communications logoLink to Nature Communications
. 2026 Jun 10;17:7830. doi: 10.1038/s41467-026-74335-y

Pharmacokinetics, bactericidal activity and toxicity of short oral regimens for rifampicin-resistant tuberculosis treatment

Bern-Thomas Nyang’wa 1,2,✉, Ilaria Motta 3, Ronelle Moodliar 4, Varvara Solodovnikova 5, Shakira Rajaram 6, Mohammed Rasool 6, Catherine Berry 7, David A J Moore 8, Geraint Davies 9, Frank Kloprogge 2,✉
PMCID: PMC13439552  PMID: 42270675

Abstract

The exposure and both Mycobacterium tuberculosis clearance rates and toxicity relationships of bedaquiline-pretomanid-linezolid- (BPaL), BPaL-clofazimine (BPaLC) and BPaL-moxifloxacin (BPaLM) for treatment of rifampicin-resistant tuberculosis remain understudied. Therefore, the relationship between the patients’ exposure to anti-TB drugs in TB-PRACTECAL trial investigational regimens and their treatment outcomes was investigated. PRACTECAL-PKPD was a prospective pharmacokinetics and pharmacodynamics study. Patients with rifampicin-resistant tuberculosis were enrolled from Belarus and South Africa. Antimicrobial exposures for bedaquiline, pretomanid, linezolid, moxifloxacin and clofazimine were adequately estimated, were within the ranges of previously published studies but did not correlate with the speed of sputum bacterial clearance. When compared to the standard of care (SoC) arm, a 20% increased bacillary killing rate with BPaLM was observed, whilst BPaL and BPaLC displayed a 15% decreased rate. Also of note was a 20% decreased bacillary killing rate in patients with severe disease irrespective of treatment. Linezolid exposure was higher amongst patients with anaemia or neutropenia. No other exposure-toxicity relationships were identified for all other drugs. These data indicate that the studied doses of linezolid and moxifloxacin could be the right balance between effectiveness and safety, strengthening the WHO recommendation that BPaLM is the preferred regimen for rifampicin-resistant tuberculosis in adolescents and adults. The study was registered in Clinical Trials.gov, TRN: NCT04081077.

Subject terms: Tuberculosis, Pharmacology, Infection

Authors present results from a prospective pharmacokinetics and pharmacodynamics study (PRACTECAL-PKPD), where they studied the relationship between patients’ exposure to anti-tuberculosis drugs, and their treatment outcomes. Antimicrobial exposures were within expected ranges but did not correlate with the speed of bacterial clearance nor treatment outcomes. Linezolid exposure was higher in anaemic and neutropenic patients, but no other exposure-toxicity relationships.

Introduction

Each year, around 500,000 patients fall ill with tuberculosis that is resistant to rifampicin, with or without additional resistance to isoniazid1. Until 2022, when WHO provisionally recommended the use of Bedaquiline-Pretomanid-Linezolid-Moxifloxacin (BPaLM) for treatment of drug-resistant TB, treatment was long at 9–20 months with considerable toxicity and poor efficacy at around 60%2. Despite the Bedaquiline–Pretomanid–Linezolid (BPaL) backbone showing superior efficacy compared to local standard of care (SoC) regimens, at reduced treatment duration with less toxicity, Mycobacterium tuberculosis (Mtb) clearance from sputum and toxicity and their correlation with antimicrobial exposure of these regimens remain understudied. A detailed understanding could facilitate further dose optimisation to maximise bactericidal activity whilst minimising toxicity.

Bedaquiline is a diarylquinoline antimycobacterial which inhibits the proton pump of mycobacterial ATP synthase3. Pretomanid is a nitroimidazooxazine that inhibits mycolic acid biosynthesis, thus disrupting cell wall production in actively replicating Mtb4; Linezolid is an oxazolidinone that inhibits ribosomal 8 protein synthesis by binding to the 23S RNA peptidyl transferase centre of the 50S subunit of the prokaryotic ribosome5; Clofazimine is a lipophilic riminophenazine that acts through intracellular redox cycling, interfering with potassium uptake in membrane phospholipids and anti-inflammatory activity through inhibition of T-lymphocytes activation and proliferation6. Moxifloxacin is a fluoroquinolone that inhibits DNA gyrase, which ultimately inhibits the replication of Mtb DNA7.

Linezolid is part of the investigational treatment backbone in TB-PRACTECAL, although there are concerns around the safety of the administered dose. The most widely used dose of 600 mg is known to be associated with toxicity, given the long duration of linezolid intake for treatment of Mtb. Dose reduction to 300 mg is therefore not uncommon and was used routinely after four months in TB-PRACTECAL8,9. All other drugs were given at the labelled doses.

Pharmacokinetic–pharmacodynamic (PK–PD) indices, describing the association between antibiotic exposure and Mtb clearance have been derived from animal models or hollow-fibre systems for tuberculosis (HFS-TB), typically under monotherapy conditions10–13. Subsequent probability of target attainment (PTA) simulations, that have informed dosing regimens, may consequently indicate suboptimal target attainment as the drug combination effect is not being accounted for. However, assuming at least additive effects, exposures may still be sufficient when drugs are administered in combination to achieve complete eradication of Mtb.

Sputum Mtb load data from patients receiving combination treatment is available, but is typically analysed as a binary variable, i.e., conversion to negative Mtb sputum after 2 or 3 months of treatment14. This approach omits data and therefore discards important information. Characterising longitudinal Mtb sputum load data using statistical modelling techniques enables summarising how sputum clearance evolves over time and enables correlating longitudinal changes in sputum Mtb load with treatment efficacy15,16.

We, therefore, aimed to study the relationship between the patients’ exposure to anti-TB drugs in the TB-PRACTECAL trial investigational regimens and their treatment outcomes.

Results

Between January 2017 and March 2021, 552 patients were enroled in the main study, of whom 94, 458 and 514 patients contributed to the pharmacokinetic, mycobacteriology and adverse event analyses. The baseline age and BMI of all three populations were similar, however the proportion with HIV was higher at 42% (39/94) for the PK population. About 359 patients contributed two or more MGIT-TTP data samples during the first 12 weeks treatment as a measure of sputum bacillary load evolution over time, 274 (76%) had smear-positive tuberculosis, and 256 (71%) had visible tuberculosis cavities on chest x-ray (Table 1).

Table 1.

Baseline characteristics of the study population

Pharmacokinetics Mycobacteriology Toxicity TB-PRACTECAL
Patients (n) 94 359 514 552
Age (years, median [IQR]) 36 [31–43] 35 [27–43] 35 [27–43] 35 [27–43]
Female (n,%) 34 [36] 147 [41] 216 [42] 223 [40]
Weight (kg, [IQR]) 56.8 [48.2–67.775] 55.9 [48.5–64.05] 56.5 [49.925–65] 56.9 [50–65.625]
BMI (kg/m2, [IQR] 19.75 [17.6–22.2] 19.2 [17.45–22] 19.8 [17.6–22.3] 19.7 [17.7–22.3]
HIV (n,[%]) 39 [41] 108 [30] 146 [28] 153 [28]
Smear 1 + (n,[%]) 9 [10] 57 [16] 69 [13] 75 [14]
Smear 2 + (n,[%]) 12 [13] 105 [29] 110 [21] 114 [21]
Smear 3 + (n,[%]) 29 [31] 112 [31] 110 [21] 118 [21]
Cavity (n,[%]) 37 [40] 256 [71] 310 [60] 326 [59]
Open in a new tab

IQR inter-quartile range, BMI body mass index.

Antimicrobial drug exposure was available for patients contributing two or more MGIT-TTP data, while all 94 patients had both drug exposure and related adverse event data (Fig. 1 and Table 1). A study profile characterising the path that the final analysis population per arm took through the clinical trial and sub-study is presented in Supplementary Fig. S1.

Fig. 1.

Fig. 1

Open in a new tab

Study profile showing the study components and the analysis populations.

For 37 patients in the BPaLM arm, 298 (median [range], 9 [4–9] samples/patient) moxifloxacin plasma concentration samples were available. Absorption, distribution and elimination of moxifloxacin in plasma was described using a two-compartment distribution model, first-order absorption and elimination (Table S1 and Fig. 2). Elimination clearance was at 14.2 l/h (%CV, 23.1) and corresponding median [IQR] steady state moxifloxacin AUC0–24 was 26.5 [20.0–40.0] mg*hr/l (Table 2).

Fig. 2. Visual predictive check of the moxifloxacin population pharmacokinetic model, showing plasma concentration vs time data.

Fig. 2

Open in a new tab

All data from the baseline and month 2 visit in the left panel are shown as time after last dose, and through-level data from the visits beyond month two in the right panel as time after the first dose. Dots represent observations, solid and dashed lines the median and 2.5–97.5 percentiles of the observed data with corresponding 90% confidence intervals overlaid (nsimulations = 2000). Source data are provided as an R model file labelled 'moxifloxacin.final.model'.

Table 2.

Individual drug exposure metrics

Pretomanid 200 mg Clofazimine 100 mg Linezolid 600 mg Moxifloxacin 400 mg Bedaquiline 200 mg*
Median Range Median Range Median Range Median Range Median Range
AUC0–24 (mg × h/L) 63.73 30.85–138.18 15.67 1.87–24.61 85.76 35.68–156.88 26.48 19.98–40.02 42.80 15.1–74.9
AUC0–48 (mg × h/L) 74.34 24.6–140
AUC0–72 (mg × h/L) 104.51 32.9–201
Ctrough (mg/L) 1.97 0.065–7.70 0.66 0.02–1.47 0.71 0.12–2.98 0.42 0.24–0.80 1.32 0.18–3.45
Cmax (mg/L) 3.19 1.38–6.35 0.71 0.12–1.1 8.28 4.48–12.96 1.99 1.43–2.64 2.46 0.88–6.15
Open in a new tab

Pretomanid, linezolid and moxifloxacin estimation was done at steady state (week 8), bedaquiline and clofazimine estimation at week 24 of treatment.

AUC area under the concentration-time curve, Ctrough concentration just before the next dose, Cmax highest concentration within a dosing interval.

*Loading dose at 400 mg daily for 2 weeks, then 200 mg x3/week.

Across the BPaL, BPaLC and BPaLM arms, a total of 94 patients were enrolled that contributed 952 (median [range], 11 [3–12] samples/patient) bedaquiline plasma concentration samples. Absorption, distribution and elimination of bedaquiline in plasma was described using a three-compartment distribution model, transit absorption and elimination (Table S1 and Fig. 3). Elimination clearance was at 1.93 l/h (%CV, 43.1) and corresponding median [range] bedaquiline AUC0–48 and AUC0–72 was 74.3 [24.6–140] and 104.5 [32.9–201] mg*hr/l at treatment completion (Table 2).

Fig. 3. Visual predictive check of the bedaquiline population pharmacokinetic model, showing plasma concentration vs time data.

Fig. 3

Open in a new tab

All data from the baseline and month 2 visit in the left panel as time after last dose, and through-level data from the visits beyond month 2 in the right panel as time after the first dose. Dots represent observations, solid and dashed lines the median and 2.5–97.5 percentiles of the observed data with corresponding 90% confidence Intervals overlaid (nsimulations = 2000). Source data are provided as an R model file labelled 'bedaquiline.final.model'.

Exposure estimates for all study drugs are summarised in Table 2. Population PK model estimates for pretomanid, linezolid and clofazimine17–19, and for both moxifloxacin and bedaquiline are summarised in Table S1.

Baseline sputum bacillary load expressed as the median MGIT time to positivity (MGIT TTP) in days was 14.4, 8.9, 6.2 and 5.1 days for patients with smear-negative, 1+, 2+ and 3+, respectively. The bacillary clearance model VPCs demonstrated their predictive accuracy (Fig. S2).

Sputum bacillary clearance was slower for patients with smear 2+ at baseline, at a MGIT TTP slope of 0.804, compared to smear 1+ or negative patients. Cavitary disease at baseline also resulted in reduced sputum bacillary clearance at an MGIT TTP slope of 0.728 compared to cavitary-negative patients. Sputum bacillary clearance in the BPaLM arm was faster compared to the SoC arm, whilst the sputum bacillary clearance in the BPaL/BPaLC arm was slower compared to the SoC arm. The MGIT-TTP slope for the BPaLM arm was 20% steeper than the SoC arm, whilst the MGIT-TTP slope in the BPaL/BPaLC arm was 15% less steep than the SoC (Fig. 4 and Table S2).

Fig. 4. Effect of baseline covariates and treatment regimens on bacillary clearance.

Fig. 4

Open in a new tab

Using model-predicted MGIT-TTP vs time on treatment profiles for six representative patients. The participants with favourable baseline covariates (negative smear and absence of lung cavitation) are shown in green, whereas the participants with unfavourable baseline covariates (smear 3+ and presence of lung cavitation) are shown in orange. The solid, dotted, and dot-dashed lines represent SoC, BPaL/BPaLC and BPaLM, respectively. The black dashed horizontal line represents a TTP of 42 days. MGIT-TTP, time to positivity. SoC standard of care, BPaL/BPaLC bedaquiline, pretomanid, linezolid plus/minus clofazimine and BPaLM bedaquiline, pretomanid, linezolid and moxifloxacin. Source data are provided as an R model file labelled 'MGIT-TTP_final_model'.

Neither bedaquiline, moxifloxacin, nor previously reported pretomanid, linezolid or clofazimine exposures correlated with random variability on sputum bacterial clearance between patients in EBE plots (Figs. 5 and 6) and when explored as a fixed effect covariate. Additionally, there were no differences in the antimicrobial exposure by the various long-term trial outcomes (Fig. S3).

Fig. 5. ndividual drug exposure-response plots of random variability on TTP (ETA) slope vs individual model predicted log AUC0–24 for bedaquiline (n = 49), pretomanid (n = 49), linezolid (n = 42), moxifloxacin (n = 19) and clofazimine (n = 18).

Fig. 5

Open in a new tab

The solid lines show a LOESS (locally estimated scatterplot smoothing) fit to the data, with the shaded area indicating the 95% confidence interval for the smooth. Source data are provided as an Excel file called 'Source Data' in a sheet labelled Fig. 5.

Fig. 6. Individual drug exposure-response plots of random variability on TTP (ETA) slope vs individual model predicted T > MIC for pretomanid (n = 57) and clofazimine (n = 21).

Fig. 6

Open in a new tab

The solid lines show a LOESS (locally estimated scatterplot smoothing) fit to the data, with the shaded area indicating the 95% confidence interval for the smooth. Source data are provided as an Excel file called 'Source Data' in a sheet labelled Fig. 6.

A total of 280 patients had anaemia, decreased creatinine clearance, elevated ALT, AST, bilirubin, neutropenia or peripheral neuropathy, ranging from grade 1–4, 89, 55, 65 and 71 patients in the SoC, BPaL, BPaLC and BPaLM arm (Fig. S4). For those with antimicrobial plasma concentrations, 38 patients had AEs, ranging between grade 1 and grade 2. In one-way ANOVA analyses, neither bedaquiline, pretomanid, moxifloxacin nor clofazimine AUC0–24 was correlated with liver dysfunction, i.e. ≥grade 1 elevated AST, ALT or bilirubin. Linezolid AUC0–24 was correlated with ≥grade 1 anaemia and neutropenia (Fig. 7). Individual bedaquiline, pretomanid, linezolid, moxifloxacin or clofazimine AUC0–24 did not correlate with the severity grade for each of the investigated adverse events, with most of the adverse events being grade 1.

Fig. 7. Exposure-toxicity analyses (n = 94) box plots of AUC0-24 distribution of bedaquiline, pretomanid, linezolid, moxifloxacin and clofazimine for each related adverse event.

Fig. 7

Open in a new tab

Boxes show the inter-quartile range (25–75th percentiles) with the median indicated by the central line; whiskers represent the 1.5x inter-quartile range. One-way ANOVA results: ns (no difference), ** Linezolid-anaemia p value = 0.00455, *** linezolid-neutropenia p value = 0.000281. Source data are provided as an Excel file called 'Source Data' in a sheet labelled Fig. 7.

Discussion

This study investigated plasma antibiotic exposure of participants in the TB-PRACTECAL trial, and its relationship with Mtb sputum clearance rates, efficacy and safety outcomes.

Moxifloxacin population pharmacokinetic properties aligned with published work, elimination clearance of 14.2 l/h is within the range of previously reported estimates at 8–18 l/h12. Bedaquiline disposition kinetics, three-compartments, and elimination clearance at 1.93 l/h, corresponded with previously reported population multi-compartmental pharmacokinetic models reporting elimination clearances at 1.50–4.54 l/h3,11,20–22. Plasma exposures for bedaquiline, pretomanid, linezolid, moxifloxacin and clofazimine were within previously published efficacious ranges17–19.

Longitudinal data analyses of sputum Mtb load confirmed BPaLM’s superior capacity to clear Mtb from sputum compared to SoC, which was previously described by time to sputum conversion. The study also showed that BPaL and BPaLC were inferior in clearing Mtb from sputum compared to SoC, likely due to the contribution of the fluoroquinolone (97%), linezolid (87%) and bedaquiline (88%) in the SoC regimens. Amongst the investigated regimens in TB-PRACTECAL, BPaLM displayed the highest bacillary clearance, and effectiveness on the primary endpoint of the trial, while the inferior bacillary clearance of BPaL and BPaLC were also reflected in slightly higher rates of microbiological treatment failure and relapse9. Although our findings show concordance between sputum bacillary clearance rates and treatment efficacy, this association may be context-specific and should not be interpreted as a universally predictive marker23.

Variability in rifampicin exposure after standard dosing has been well known to correlate with sputum Mtb clearance during first-line combination therapy of TB, highlighting that the pharmacodynamic effect is not saturated, supporting endeavours to increase the dose of rifampicin15,24. In this study, associations between Mtb sputum clearance and plasma antibiotic exposure were absent, which could suggest that current doses achieve plasma levels sufficiently high enough to saturate Mtb clearance from sputum. However, the PKPD dataset is small and may not have had sufficient power to detect differences among drugs with correlated exposures in the context of combination therapy with fixed dose levels. Furthermore, highly bactericidal drugs such as moxifloxacin may have similar effects to isoniazid, in masking the dose-response effect of accompanying drugs25.

A trend of higher linezolid AUC0-24 was observed in patients with anaemia, neurotoxicity and peripheral neuropathy compared to those patients without. However, the linezolid-containing investigational arms in the TB-PRACTECAL study were generally well tolerated, with only grade 1 and 2 adverse events being observed amongst patients that also had linezolid concentrations quantified in plasma. This confirms previously reported associations between linezolid plasma exposure and toxicity, while it also underpins the success of a step down from daily linezolid at 600 to 300 mg after 16 weeks, to prevent toxicity8.

The main limitations of this study derive from its restricted participant recruitment due to it being a sub-study: as patients with moderate and severe renal, hepatic, haematologic or cardiac disease were excluded from the RCT, the results may not be representative of the full spectrum of RR-TB patients. In addition, the study recruited 94 of the 240 participants assumed in the optimal design process, which limited the exposure variability and may be one of the explanations for the absence of the exposure-effect association. Finally, although the optimal design analyses indicated that the timing of the samples were adequate, the blood sampling schedule was originally optimised for linezolid pharmacokinetics only.

Though PRACTECAL-PKPD was not a dose-ranging trial, these data on early Mtb clearance from sputum and adverse events, taken in conjunction with the primary analysis, indicate that the studied doses of linezolid and moxifloxacin are effective and safe. This strengthens the WHO recommendation to include BPaLM as the preferred drug combination for treatment of pulmonary rifampicin-resistant tuberculosis in adolescents and adults.

Methods

Ethical approval was obtained from the London School of Hygiene & Tropical Medicine Research Ethics Committee and the Médecins Sans Frontières Ethics Review Board. Site-specific approvals were obtained from the Belarus Republican Scientific and Practical Centre of Pulmonology and Tuberculosis ethics committee for the Minsk site, PharmaEthics for the Don Mckenzie and Dorris Goodwin hospitals sites and the University of Witwatersrand Human Research ethics committee for the Helen Joseph and King DiniZulu Hospitals sites. Written informed consent was obtained from all participants before recruitment into the TB-PRACTECAL trial, and sub-study-specific consent was obtained before any data or sample collection for the PRACTECAL-PKPD. All clinical activities were conducted in accordance with the International Council for Harmonisation’s (ICH) Good Clinical Practice, and laboratory investigations were conducted in accordance with Good Clinical and Laboratory Practice.

Study design

PRACTECAL-PKPD (NCT04081077) was a prospective optimal design-based pharmacokinetics and pharmacodynamics study nested in TB-PRACTECAL (NCT02589782)—an open-label, randomised, controlled, multi-arm, phase IIb-III, multicentre, non-inferiority trial which investigated bedaquiline-pretomanid-linezolid (BPaL), BPaL-Clofazimine (BPaLC), and BPaL-Moxifloxacin (BPaLM) compared to standard of care for patients with rifampicin-resistant pulmonary tuberculosis26.

Participants

Patients aged 15 years or older with rifampicin-resistant pulmonary tuberculosis enrolled in the investigational arms of TB-PRACTECAL at five hospitals and health centres in Belarus and South Africa were offered enrolment into the PKPD study.

The key inclusion and exclusion criteria were: adolescent or adult with pulmonary rifampicin-resistant tuberculosis, irrespective of HIV infection or CD4 count. And exclusion of those who were pregnant, with alanine aminotransferase (ALT) concentration or an aspartate aminotransferase (AST) concentration higher than three times the upper limit of the normal range, any baseline laboratory value consistent with Grade 4 toxicity or a Fridericia-corrected QT (QTcF) interval longer than 450 ms26,27.

Procedures

BPaLM, BPaLC and BPaL were taken orally for 24 weeks. Bedaquiline was administered at 400 mg daily for 2 weeks, followed by 200 mg three times weekly for the remaining 22 weeks, pretomanid 200 mg daily for 24 weeks, and linezolid 600 mg daily for the first 16 weeks, followed by 300 mg for the remaining 8 weeks. BPaLM additionally contained moxifloxacin 400 mg daily for 24 weeks and BPaLC, clofazimine 100 mg daily (or 50 mg for patients weighing <33 kg) for 24 weeks. Comparator treatment comprised locally accepted standard-of-care (SoC), that was updated in line with international recommendations over the course of the study; in summary, the standardised 9–11 months regimens in Belarus consisted of high-dose isoniazid, ethambutol, pyrazinamide, capreomycin, moxifloxacin, ethionamide and clofazimine (Hh/E/Z/Cm/Mfx/Eto/Cfz). In South Africa, this was later replaced by bedaquiline, levofloxacin, linezolid, clofazimine and terizidone (B/Lfx/Lzd/Cfz/Trd). The longer 18–20 months regimen consisted of Z, Cm, Mfx, B, Lzd plus Eto, terizidone or PAS (Z/Cm/Mfx/B/Lzd plus Eto, Trd or PAS) and was individualised. From 2017 to 2019, these regimens generally included a second-line injectable agent, and the criteria for including bedaquiline were stringent. From 2019 to 2022, participants received all-oral versions of these regimens, and most regimens included bedaquiline. Study treatments were delivered either through direct-observed therapy or video-observed therapy. The treatment was administered and observed daily, 7 days a week in the investigational arms and at least 6 days a week in the SOC26,28.

For the pharmacokinetic analyses, plasma samples were collected at zero, two and 23 h post dose during the baseline visit, at zero, 6.5 and 23 h post dose during the week 8 visit, followed by trough samples at the week 12, 16, 20 and 24 visits and additional samples at week 32 and week 72. Drugs were quantified using high-performance liquid chromatography-tandem mass spectrometry27.

Participants submitted spot and early morning sputum samples at day 0, day 7 and weeks 4, 8, 12, 16, 20 and 24 during treatment for microbiological analyses. Time to positivity in primary culture and minimum inhibitory concentrations were determined in the BD BACTEC™ Mycobacterial Growth Indicator Tube system (MGIT™) (Becton & Dickinson (BD)29.

Safety monitoring was conducted at least every 4 weeks and included blood chemistry analysis and physical examinations26. Adverse events (AEs) were either spontaneously reported or elicited during open-ended questioning, examination, or evaluation of a trial participant. Each AE was then evaluated to determine the severity grade: Severity grading 1, 2, 3 or 4, was based on the MSF severity grading scale, which is based on the standardised and commonly used toxicity table for infectious diseases, the Division of Microbiology and Infectious Diseases (DMID) grading system, complemented with a selection of terms from the NCI’s Common Terminology Criteria for Adverse Events (CTCAE) scale.

Outcomes

The first objective was to develop drug exposure metrics for bedaquiline, pretomanid, linezolid, moxifloxacin and clofazimine. The key model-derived outcomes were the area under the drug concentration curve (AUC), the lowest drug concentration just before the next dose (Ctrough), the highest drug concentration between two doses (Cmax). Together with the cumulative percentage of the dosing interval that the drug concentration exceeds the MIC (T > MIC), drug exposure matrices were evaluated, adjusted for MIC or not, in the context of bacillary clearance.

The efficacy (second and third) objectives were to establish an exposure-response model for each drug and regimen to both early bactericidal activity and long-term treatment outcomes. The main endpoint was the change over time to the duration it took for a culture to have growth detected in MGIT, i.e., time to positivity (TTP).

The safety (fourth) objective was to investigate the exposure-toxicity relationship of each drug. The outcome measures for this analysis were adverse events, including elevated ALT, AST or bilirubin levels, decreased creatinine clearance, anaemia, neutropenia and peripheral neuropathy.

Data collection and analysis software

Clinical data, including patient baseline characteristics, adverse events and treatment outcomes, were collected in an electronic clinical report form (eCRF) using OpenClinica version 3.12, processed and analysed using Stata statistical software version 15.1. Bioanalytical data acquisition and integration was performed by Analyst version 1.6.1 and Multi Quant version 3.0. Mycobacteriology data were collected using the MGIT 960 instrument connected to an EpiCenter™ equipped with the TBeXIST module (BD). Dose intake and sample collection timing data were collected using KoboCollect version 1.14.0a. Analyses including nonlinear mixed-effect modelling were performed in R v4.1.2 using the nlmixr2 package for the population PK and pharmacodynamic model development.

Statistics and reproducibility

No statistical method was used to predetermine sample size, however, during the optimal design process, we used the Fisher Information Matrix to optimise a venous blood sampling design that supports pharmacokinetic model development, which achieved the objectives at a sample size of 24027. Patients in the parent trial were randomly assigned in a 1:1:1:1 ratio in stage 1 to BPaLM, BPaLC, BPaL and SoC; and in a 1:1 ratio to BPaLM and SoC in stage 2. Randomisation lists were prepared by trial statisticians, and randomisation was stratified by study site. The Investigators were not blinded to allocation, however the mycobacteriology laboratory technicians and the statistician for the outcome analyses were26.

Bioanalytical methods were successfully validated by determination of the following parameters: specificity and selectivity; recovery; stability; matrix effect; linearity; accuracy and precision; levels of detection and levels of quantification.

Modelling methods

Pretomanid, linezolid and clofazimine AUC0–24 have been previously published17–19, whilst moxifloxacin and bedaquiline AUC0-24 were derived from moxifloxacin and bedaquiline population pharmacokinetic models described here. Moxifloxacin and bedaquiline absorption, distribution, metabolism and excretion were characterised using a population pharmacokinetic model. Due to a sparse pharmacokinetic sampling design, fixed-effects for absorption rate constant (ka), inter-compartmental clearance (q), and peripheral volume of distribution (Vp) for moxifloxacin were fixed to literature values whilst for bedaquiline only ka was fixed to literature values12,20. Fat-free mass (FFM) and body mass index (BMI) were embedded in the model a priori based on previous literature reports20,30. Antimicrobial exposure indices, i.e. area under the concentration-time curve (AUC0–24)/MIC and T > MIC, were investigated as explanatory variables for variability in MGIT TTP slope over time.

MGIT TTP-TIME data from the first twelve weeks of treatment in patients contributing at least two or more MGIT TTP samples was retained for data analysis, as later positive samples represented relapses. A linear, exponential and power equation with random effects representing between-patient and residual variability, were fitted to the data and compared for superiority using the Bayesian Information Criterion. MGIT-TTP values above the limit of quantification were handled with the censoring method in nlmixr2, based on the M3 methods in NONMEM. The upper limit of quantification was set to 42 days in line with the threshold for an MGIT-TTP negative sample. Clinically relevant explanatory variables were tested as proportional covariates on baseline MGIT TTP and MGIT TTP slope over time (p < 0.01 for 1 dF); these included baseline cavitation, smear status, HIV status, age and study arm, and these were candidate explanatory variables for variability in baseline MGIT TTP and slope over time.

The relationship between elevated AST, ALT, and bilirubin, decreased creatinine clearance, anaemia, neutropenia and peripheral neuropathy with AUC0-24 was analysed using one-way ANOVA.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information (571.5KB, pdf)
Reporting summary (92.2KB, pdf)
Transparent Peer Review file (268.9KB, pdf)

Source data

Source Data (875.1KB, zip)

Acknowledgements

We thank all TB-PRACTECAL clinical trial participants for volunteering to be part of this study.

Author contributions

Conception and design: B.-T.N., F.K., G.D. and D.A.J.M. Data acquisition: I.M., R.M., V.S. and S.R. Pharmacokinetic modelling and first draft of manuscript: B.T.-N. and F.K. Revising and approval of manuscript: B.-T.N., I.M., R.M., V.S., S.R., M.R., C.B., D.A.J.M., F.K. and G.D.

Peer review

Peer review information

Nature Communications thanks Shashikant Srivastava, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

Médecins sans Frontières was the funder and sponsor of the study and was involved in study design, data analysis, data interpretation, and writing of the report. B.-T.N., I.M. and C.B. were employees of Médecins Sans Frontières (MSF) at the time of data collection and analysis. F.K. is a recipient of a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (Grant Number 220587/Z/20/Z). R.M., V.S., S.R., M.R., D.A.J.M. and G.D. declared no relevant funding.

Data availability

Deidentified pharmacokinetic data will be available to researchers upon a written request to the Medical Director, Médecins sans Frontières, Operational Centre Amsterdam, the Netherlands. Clinical and mycobacteriology data were available from the TB-PACTS study ID 1035 at: https://c-path.org/tools-platforms/tb-pacts and the World Health Organization Tuberculosis treatment individual patient data platform, University College London. Available at: https://www.ucl.ac.uk/global-health/research/z-research/tb-ipd-platform. Source data are provided with this paper.

Competing interests

B.-T.N., I.M. and C.B. declare employment at Médecins Sans Frontières (MSF), the sponsor of the TB-PRACTECAL trial, while serving as the Chief Investigator, Medical Monitor and Global Principal Investigator of the TB-PRACTECAL trial, respectively. The remaining authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors jointly supervised this work: Geraint Davies, Frank Kloprogge.

Contributor Information

Bern-Thomas Nyang’wa, Email: bern.nyangwa@london.msf.org.

Frank Kloprogge, Email: f.kloprogge@ucl.ac.uk.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74335-y.

References

  • 1.Global tuberculosis report 2025. (Geneva: World Health Organization; 2025). Licence: CC BY-NC-SA 3.0 IGO.
  • 2.WHO. WHO Consolidated Guidelines on Tuberculosis. Module 4: Treatment (World Health Organisation, 2022).
  • 3.McLeay, S. C., Vis, P., van Heeswijk, R. P. & Green, B. Population pharmacokinetics of bedaquiline (TMC207), a novel antituberculosis drug. Antimicrob. Agents Chemother.58, 5315–5324 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Singh, R. et al. PA-824 kills nonreplicating Mycobacterium tuberculosis by intracellular NO release. Science322, 1392–1395 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Leach, K. L. et al. The site of action of oxazolidinone antibiotics in living bacteria and in human mitochondria. Mol. Cell26, 393–402 (2007). [DOI] [PubMed] [Google Scholar]
  • 6.Levy, L. Pharmacologic studies of clofazimine. Am. J. Trop. Med. Hyg.23, 1097–1109 (1974). [DOI] [PubMed] [Google Scholar]
  • 7.Gillespie, S. H. & Billington, O. Activity of moxifloxacin against mycobacteria. J. Antimicrob. Chemother.44, 393–395 (1999). [DOI] [PubMed] [Google Scholar]
  • 8.Kwak, N. et al. Optimal dosing and duration of linezolid for the treatment of multidrug-resistant and rifampicin-resistant tuberculosis: an individual patient data meta-analysis. Eur. Respir. J.66, 2500315 (2025). [DOI] [PMC free article] [PubMed]
  • 9.Nyang’wa, B. T. et al. Short oral regimens for pulmonary rifampicin-resistant tuberculosis (TB-PRACTECAL): an open-label, randomised, controlled, phase 2B-3, multi-arm, multicentre, non-inferiority trial. Lancet Respir. Med.12, 117–128 (2024). [DOI] [PubMed] [Google Scholar]
  • 10.Srivastava, S. et al. Linezolid dose that maximizes sterilizing effect while minimizing toxicity and resistance emergence for tuberculosis. Antimicrob. Agents Chemother.61, e00751-17 (2017). [DOI] [PMC free article] [PubMed]
  • 11.Shao, G. et al. Population pharmacokinetics and model-based dosing evaluation of bedaquiline in multidrug-resistant tuberculosis patients. Front. Pharm.14, 1022090 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Zvada, S. P. et al. Moxifloxacin population pharmacokinetics and model-based comparison of efficacy between moxifloxacin and ofloxacin in African patients. Antimicrob. Agents Chemother.58, 503–510 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Ahmad, Z. et al. PA-824 exhibits time-dependent activity in a murine model of tuberculosis. Antimicrob. Agents Chemother.55, 239–245 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Conradie, F. et al. Treatment of highly drug-resistant pulmonary tuberculosis. N. Engl. J. Med.382, 893–902 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Kloprogge, F. et al. Longitudinal pharmacokinetic-pharmacodynamic biomarkers correlate with treatment outcome in drug-sensitive pulmonary tuberculosis: a population pharmacokinetic-pharmacodynamic analysis. Open Forum Infect. Dis.7, ofaa218 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Sloan, D. J. et al. Pharmacodynamic modeling of bacillary elimination rates and detection of bacterial lipid bodies in sputum to predict and understand outcomes in treatment of pulmonary tuberculosis. Clin. Infect. Dis.61, 1–8 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Nyang’wa, B.-T. et al. Linezolid population pharmacokinetics in rifampicin-resistant tuberculosis patients and interpretation challenges of the probability of target attainment. Preprint at medRxiv10.64898/2025.12.04.25340900 (2025).
  • 18.Nyang’wa, B. T. et al. Population pharmacokinetics and target attainment of pretomanid in rifampicin-resistant tuberculosis patients. Sci. Rep.16, 15255 (2026). [DOI] [PMC free article] [PubMed]
  • 19.Nyang’wa, B.-T. et al. Clofazimine population pharmacokinetics and target attainment in rifampicin-resistant Tuberculosis patients. Preprint at medRxiv10.64898/2025.12.04.25340990 (2025).
  • 20.Svensson, E. M., Dosne, A. G. & Karlsson, M. O. Population pharmacokinetics of bedaquiline and metabolite M2 in patients with drug-resistant tuberculosis: the effect of time-varying weight and albumin. CPT Pharmacomet. Syst. Pharm.5, 682–691 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Zou, J. et al. Population pharmacokinetic modeling of bedaquiline among multidrug-resistant pulmonary tuberculosis patients from China. Antimicrob. Agents Chemother.66, e0081122 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Zhu, H. et al. Population pharmacokinetics of bedaquiline in patients with drug-resistant TB. Int. J. Tuberc. Lung Dis.25, 1006–1012 (2021). [DOI] [PubMed] [Google Scholar]
  • 23.Gillespie, S. H. et al. Four-month moxifloxacin-based regimens for drug-sensitive tuberculosis. N. Engl. J. Med.371, 1577–1587 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Te Brake, L. H. M. et al. Increased bactericidal activity but dose-limiting intolerability at 50mg·kg(-1) rifampicin. Eur. Respir. J. 58, 2000955 (2021). [DOI] [PMC free article] [PubMed]
  • 25.O’Brien, R. J. Studies of the early bactericidal activity of new drugs for tuberculosis: a help or a hindrance to antituberculosis drug development?. Am. J. Respir. Crit. Care Med.166, 3–4 (2002). [DOI] [PubMed] [Google Scholar]
  • 26.Berry, C. et al. TB-PRACTECAL: study protocol for a randomised, controlled, open-label, phase II-III trial to evaluate the safety and efficacy of regimens containing bedaquiline and pretomanid for the treatment of adult patients with pulmonary multidrug-resistant tuberculosis. Trials23, 484 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Nyang’wa, B. T. et al. Population pharmacokinetics and pharmacodynamics of investigational regimens’ drugs in the TB-PRACTECAL clinical trial (the PRACTECAL-PKPD study): a prospective nested study protocol in a randomised controlled trial. BMJ Open11, e047185 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Nyang’wa, B. T. et al. A 24-week, all-oral regimen for rifampin-resistant tuberculosis. N. Engl. J. Med.387, 2331–2343 (2022). [DOI] [PubMed] [Google Scholar]
  • 29.Rofael, S. et al. M. tuberculosis baseline and acquired drug resistance variants in TB-PRACTECAL clinical trial. Preprint at SSRN10.2139/ssrn.5941233 (2025).
  • 30.Anderson, B. J. & Holford, N. H. Mechanism-based concepts of size and maturity in pharmacokinetics. Annu. Rev. Pharm. Toxicol.48, 303–332 (2008). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information (571.5KB, pdf)
Reporting summary (92.2KB, pdf)
Transparent Peer Review file (268.9KB, pdf)
Source Data (875.1KB, zip)

Data Availability Statement

Deidentified pharmacokinetic data will be available to researchers upon a written request to the Medical Director, Médecins sans Frontières, Operational Centre Amsterdam, the Netherlands. Clinical and mycobacteriology data were available from the TB-PACTS study ID 1035 at: https://c-path.org/tools-platforms/tb-pacts and the World Health Organization Tuberculosis treatment individual patient data platform, University College London. Available at: https://www.ucl.ac.uk/global-health/research/z-research/tb-ipd-platform. Source data are provided with this paper.

Articles from Nature Communications are provided here courtesy of Nature Publishing Group

RESOURCES