Spatially and temporally resolved diffusion kinetics of bedaquiline and TBAJ-587 in hard-to-treat sites of TB disease
Jacqueline Ernest (1), Marjorie Imperial (1), Firat Kaya (2), Véronique Dartois (2), Radojka Savic (1)
(1) Department of Bioengineering and Therapeutic Sciences, University of California, San Francisco; (2) Center for Discovery and Innovation, Hackensack Meridian Health, Nutley, NJ
Tuberculosis (TB) is an infectious disease that can lead to lung cavitation and death. In the hardest-to-treat population, those with cavitary disease, highly penetrant, lipophilic drugs are needed to reach bacteria. At the same time, faster clearance is needed to avoid long windows of monotherapy that lead to resistance. Bedaquiline (BDQ) and TBAJ-587 are leading candidates for Phase II trials in drug susceptible TB. While they are lipophilic, it is unknown how quickly they can penetrate the center of the bacteria-containing lesions. In this work, we characterized accumulation into granulomas in rabbits as a function of distance from the cellular rim and time from treatment initiation. A new method, laser capture microdissection (LCM) of lesions into concentric sections, allowed, for the first time, a fully quantitative investigation of the spatial drug gradients from the cellular rim to the center of caseum (cheese-like substance within necrotic cores). We developed a multicompartment site-of-action model and performed simulations to determine lesion temporal and spatial coverage and drug residence after the end of treatment.
Two studies were investigated in TB-infected rabbits: 1.) 400 mg and 300 mg for BDQ and TBAJ-587, respectively, dosed for three days, 2.) 80 mg dosed for 1, 17, 26, 28 days for BDQ and 30 mg for 14 and 28 days for TBAJ-587. BDQ, TBAJ-587, and their active metabolites (BDQ-M2 and TBAJ-587-M3) were quantified. For caseum samples, the average distance from the caseum border in micrometers (um) was calculated by creating a caseum ‘mask’ and processing it using the Exact Euclidean Distance Transform plugin of the ImageJ software.
The modeling analysis was performed in NONMEM 7.4.2. A site-of-action model was developed where a rate (klesion) and extent of partitioning (PClesion) were estimated (dClesion /dt = klesion * (PClesion * Cplasma – Clesion) . Distance from the cellular rim was tested as a covariate on PClesion and klesion. Multiple covariate relationships were tested including linear, loglinear, power, sigmoidal Emax, and inverse relationship. To predict patient lesion coverage, the lesion parameters were linked to a clinical PK model for BDQ  and allometric scaled for TBAJ-587. Simulations were compared to potency metrics in an ex vivo caseum assay (casMBC90, minimum concentration to kill 90% bacteria) . Lesion coverage (PK > casMBC90) as a function of distance was quantified.
Overall, 55 rabbits contributed 464 measurements from plasma and 140 from lesion for BDQ and BDQ-M2. 35 rabbits contributed 544 measurements from plasma and 155 from lesion for TBAJ-587 and TBAJ-587-M3. Plasma PK were described by an oral two-compartment distribution model.
BDQ concentrations in caseum were greatest near the edge of the rim. As dosing continued, the gradient flattened across the distance. The inverse covariate on distance best described the relationship and was used for simulations. For BDQ standard regimen, caseum concentrations were above the casMBC90 after 2 weeks of dosing at the caseum edge (<300 um). At the end of treatment, coverage was achieved at distances up to 1800 um from the cellular rim. BDQ remained in caseum for over two years after treatment end. BDQ-M2 had a slightly faster rate of penetration but a lower extent.
Steeper gradients were observed for TBAJ-587 than for BDQ, yet was more potent in the caseum assay. Simulations of 400 mg QD produced TBAJ-587 concentrations above casMBC90 as far as 600 um. TBAJ-587-M3 metabolite present at effective concentrations up to 300 um. The concentration remained over the casMBC90 in caseum edges (<300 um) for about 3 weeks post-treatment and over the limit of quantification (1 ng/mL) for 7 weeks post-treatment.
A spatially resolved PK model that captures the movement of drug from plasma to centers of necrotic cores in hard-to-treat TB lesions was developed. BDQ exhibits sustained drug residence time in caseum beyond 2 years after end of treatment, which creates prolonged periods of monotherapy that may trigger resistance development. TBAJ-587 has effective concentration without extended residence time after treatment which may reduce the risk of resistance development compared to BDQ. The novel drug quantification method and novel spatiotemporal PK modeling offers a tool to assess drug penetration in hard-to-reach sites of disease, to select promising regimens, and to predict Phase III trial success.
1. Kjellsson, M.C., et al., Pharmacokinetic evaluation of the penetration of antituberculosis agents in rabbit pulmonary lesions. Antimicrob Agents Chemother, 2012. 56(1): p. 446-57. 2. Svensson, E.M., A.G. Dosne, and M.O. Karlsson, Population Pharmacokinetics of Bedaquiline and Metabolite M2 in Patients With Drug-Resistant Tuberculosis: The Effect of Time-Varying Weight and Albumin. CPT Pharmacometrics Syst Pharmacol, 2016. 5(12): p. 682-691. 3. Sarathy, J.P., et al., Extreme Drug Tolerance of Mycobacterium tuberculosis in Caseum. Antimicrob Agents Chemother, 2018. 62(2).
We would like to acknowledge the TB Alliance for providing TBAJ-587, and the Bill and Melinda Gates Foundation for funding of this work under the grant (INV-0002483). We would like to acknowledge the National Institute of Health for contributing data to produce the rabbit BDQ plasma model.