Central nervous system antituberculosis drug exposures in a rabbit model of tuberculous meningitis

Treatment optimization through pharmacokinetic evaluation is needed to improve outcomes in tuberculous meningitis (TBM). Clinical assessment of drug penetration into the central nervous system (CNS) is limited to cerebrospinal fluid (CSF) sampling, which may not reflect drug exposures at the site-of-disease. We characterized the pharmacokinetics of antituberculosis drugs in CNS tissues using a rabbit model of TBM. Rabbits infected with Mycobacterium tuberculosis received human-equivalent doses of antituberculosis drugs after displaying signs of TBM. Blood was intensively sampled on day 1 and until euthanasia on day 3 when terminal plasma, CSF, and CNS tissues were collected. Total drug concentrations were quantified by LC-MS/MS and analyzed in NONMEM to estimate equilibration half-lives and CSF- and CNS tissue-to-plasma ratios. Across 5 experiments, 48 rabbits provided 600 plasma samples and 48 sets of terminal CSF/CNS tissue samples. Pyrazinamide and isoniazid showed high and consistent penetration, with CSF ratios of 0.98 and 0.85 and CNS tissue ratios ranging from 0.44 to 0.73 and 0.60-0.82, respectively. Linezolid demonstrated modest and uniform penetration (CSF ratio 0.39; CNS tissue ratios 0.15-0.44). Rifampicin showed the lowest overall penetration (CSF ratio 0.08; CNS tissue ratios 0.07-0.32). In contrast, the most lipophilic drugs, bedaquiline and rifabutin, showed low CSF ratios (0.01 and 0.42, respectively) compared with much higher CNS tissue ratios (1.11-6.34 and 1.23-6.78, respectively). CSF penetration in the rabbit TBM model closely mirrored reported human data, establishing its translational value. High relative CNS exposures of bedaquiline and rifabutin support their clinical evaluation for TBM treatment.