henhaobio
Sign in

The Journal of Infectious DiseasesPharmacology & ToxicologyPeer-reviewed journal article

The kinetics of bedaquiline diffusion in tuberculous cavities opens a window for emergence of resistance.

Annamarie E. BustioniD, Jacqueline P. ErnestiD, Firat KayaiD, Connie Silva, Jansy Passiflora SarathyiD, Landry BlanciD, Marjorie Z. ImperialiD, Martin GengenbacheriD, Min XieiD, Matthew D. ZimmermaniD, Gregory T. RobertsoniD, Danielle Weiner, Laura E. ViaiD, Clifton E. BarryiD, Radojka M. SavićiD, Veronique Anne DartoisiD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Cavitary tuberculosis (TB) is difficult to cure and a site of relapse. Bedaquiline has been a wonder drug in the treatment of multidrug resistant TB, but emergence of resistance threatens its sustained success. To investigate the role of drug distribution in resistance development, we designed a novel laser-capture microdissection scheme to spatially resolve the penetration of bedaquiline in the necrotic center (caseum) of cavities, a recalcitrant site of disease. Working with preclinical models that form large necrotic lesions, we profiled bedaquiline and two next generation diarylquinolines TBAJ-587 and TBAJ-876. Drug concentrations were measured in laser-captured areas of cavity caseum as a function of time and distance from blood supply. To simulate drug coverage in patient cavities, the data were modeled, and drug penetration parameter estimates were linked to clinical plasma pharmacokinetics for bedaquiline and the new diarylquinolines. Pharmacokinetic-pharmacodynamic (PK-PD) simulations revealed that bedaquiline reaches efficacious concentrations in outer and deep caseum after several weeks to months and lingers at subtherapeutic concentrations up to 3 years after therapy ends. TBAJ-587 and TBAJ-876, currently in clinical development, achieve bactericidal concentrations in caseum more rapidly and shorten the window of suboptimal concentrations post treatment compared to bedaquiline. Simulations of clinically plausible dosing schemes were conducted to guide the design of clinical trials for cavitary TB and help mitigate resistance development. In summary, the slow kinetics of diffusion of bedaquiline into and out of cavity caseum creates spatio-temporal windows of subtherapeutic concentrations. Site-of-disease simulations of TBAJ-587 and TBAJ-876 predict reduced opportunities for resistance development. SIGNIFICANCEClinical resistance to bedaquiline has emerged faster than anticipated. Understanding potential contributing factors could help curb further resistance development, not only for bedaquiline but also for the next generation diarylquinolines currently in phase 2, TBAJ-587 and TBAJ-876. Here we hypothesized and confirmed that the extended time to reach steady state and slow clearance of bedaquiline leads to extremely slow diffusion into and out of cavity caseum, a recalcitrant site of TB disease and relapse. Through modeling of experimental data in a preclinical model of cavitary TB and clinical simulations, we show that the next generation diarylquinolines may reduce spatio-temporal windows of resistance development compared to bedaquiline. Our results can inform dosing schemes of diarylquinoline-based therapies that limit resistance development.

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (14 Oct 2024) · CC BY 4.0

Rabbit lesion microdissection and pharmacokinetic modeling show bedaquiline takes months to reach deep caseum and years to clear, while TBAJ-587 and TBAJ-876 shorten these windows.

Clinical bedaquiline resistance has emerged faster than expected, and its slow plasma accumulation and clearance may leave caseum, a major relapse site, at subtherapeutic concentrations. This work quantifies drug penetration in space and time in cavitary lesions and compares bedaquiline with two next-generation diarylquinolines in clinical development.

Key results

  1. ShownCaseum concentration gradients flatten with more doses in rabbits for all three diarylquinolines. BDQ showed the same inverse trend in mice and marmosets, so several weeks of dosing are needed to approach steady state in deep caseum.

    Steep gradients after 1 and 3 doses became shallower at 14 and 28 doses in rabbits; BDQ only showed the same inverse trend in C3HeB/FeJ mice (1 and 17 days, 25 mg/kg) and marmosets (31 and 36 days, 20 mg/kg). · Slow kinetics of diarylquinoline diffusion into…

  2. SuggestedIn simulated patients, the TBAJs reach bactericidal caseum concentrations at the caseum edge faster than bedaquiline, which leaves an early window of subtherapeutic exposure.

    Simulation of 500 virtual patients: casMBC90 at 300 μm reached at one week (BDQ), two days (TBAJ-587), one day (TBAJ-876), ten days (BDQ-M2), one month (TBAJ-876-M3); TBAJ-587-M3 not within the first month. · PK-PD simulations of diarylquinolines in caseous…

  3. SuggestedBedaquiline and its metabolite persist in caseum at subtherapeutic levels far longer after therapy ends than the TBAJs, creating a prolonged window for resistance selection.

    Simulated concentrations fell below LLOQ at all caseum depths 72 months after therapy for BDQ vs 38 months (TBAJ-587) and 23 months (TBAJ-876); TBAJ-876-M3 cleared in 12 months. · PK-PD simulations of diarylquinolines in caseous…

  4. SuggestedTBAJ-587 gives the deepest bactericidal coverage of caseum, and a bedaquiline loading dose does not improve coverage over daily dosing. Bedaquiline shows a sub-therapeutic gradient into deep caseum.

    TBAJ-587 above casMBC90 up to 1800 μm; BDQ subtherapeutic gradient up to 3000 μm (maximum simulated depth); highest TBAJ doses (200 mg, 100 mg QD) reach casMBC90 at ∼1000 μm; loading dose had no significant impact. · PK-PD simulations of diarylquinolines in caseous…; Figure legends

  5. ShownA plasma-to-lesion model with a distance-dependent partition coefficient describes drug penetration into caseum. Bedaquiline has the steepest gradient, consistent with its lower free fraction and higher hydrophobicity.

    Oral two-compartment plasma model with 1 or 2 transit compartments; TBAJ-876 clearance followed Michaelis Menten kinetics; TBAJ-587 and TBAJ-876 had 10-fold higher free fraction than BDQ; VPC (N=1000) supported the fit. · Quantitative modeling of diarylquinoline…; Modeling of tissue distribution and PK-PD…

Methods, data and reproducibility

Drug concentrations in caseum were measured by laser-capture microdissection with LC/MS-MS in infected animals, then fitted with a plasma-to-lesion PK model built in NONMEM. The model was linked to clinical plasma PK models to simulate patient target attainment against ex vivo caseum MBC values.

  • Dose-finding PK studies in naive rabbits at 20, 80 and 160 mg/kg for 1 to 17 days · rabbit
  • Mtb HN878-infected rabbits with cavitary lesions, given 1 to 28 doses of BDQ, TBAJ-587 or TBAJ-876, followed by LCM of caseum · rabbit (New Zealand White)
  • TB-infected C3HeB/FeJ mice (Erdman) dosed with BDQ at 25 mg/kg for 1 and 17 days · mouse
  • TB-infected common marmosets (H37Rv) dosed with BDQ at 20 mg/kg for 31 and 36 days · common marmoset
  • Ex vivo caseum bactericidal assay on non-replicating persisters at 7 and 14 days (casMBC10, 50, 90) · Mycobacterium tuberculosis
  • Virtual patient simulations per dosing regimen of BDQ, TBAJ-587 and TBAJ-876 · human (simulated) · n = 500 virtual patients per regimen
Methods & tools
Laser-capture microdissection, LC/MS-MS, ImageJ Exact Euclidean Distance Transform, NONMEM 7.5.0, PsN 5.3.0, R 4.2.2, ggplot2, xpose, xpose4, vpc, dr4pl 2.0.0, Visual predictive checks
Reagents & environment
NONMEM 7.5.0, PsN 5.3.0, R 4.2.2 (ggplot2, xpose, xpose4, vpc), dr4pl version 2.0.0.
Reproducibility gaps
No code or data repository URLs are given in the text; assay and tissue details are deferred to Supporting Information and Tables S1 to S2.
Limitations (5)
  • The study design could not assess how lesion healing changes drug penetration over time, including after treatment ends. · stated by the authors
  • Lower blood flow at the lung apex, where relapse-prone cavities sit, is not included in the model and should be integrated in future work. · stated by the authors
  • Differences in species, pathology, sampling and modeling approach preclude direct comparison with earlier mouse-based site-of-action models. · stated by the authors
  • Patient predictions extrapolate rabbit lesion parameters onto clinical plasma PK models without direct validation against human caseum drug measurements. · noted in this summary
  • Resistance emergence is inferred from simulated durations of subtherapeutic exposure; no resistant-mutant selection was measured in this work. · noted in this summary

Read this if you model site-of-disease pharmacokinetics in TB lesions or evaluate diarylquinoline dosing for cavitary disease and resistance risk.

Citations
10
FWCI
6.68
Field percentile
top 2%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISXu '20Aguilar-Aya '26Ghodousi '19Ismail '25Kaushik '19Wetzel '26
74% semantic matchMicrobiology
J. Xu, P. J. Converse, A. Upton+3 more·bioRxiv·Nov 17, 2020