Linezolid is widely used for drug-resistant tuberculosis (DR-TB) but has a narrow therapeutic index. To inform dose optimization, we aimed to characterize the population pharmacokinetics of linezolid in South African participants with DR-TB and explore the effect of covariates, including HIV coinfection, on drug exposure. Data were obtained from pharmacokinetic substudies in a randomized controlled trial and an observational cohort study, both of which enrolled adults with drug-resistant pulmonary tuberculosis. Participants underwent intensive and sparse plasma sampling. We analyzed linezolid concentration data using nonlinear mixed-effects modeling and performed simulations to estimate attainment of putative efficacy and toxicity targets. A total of 124 participants provided 444 plasma samples; 116 were on the standard daily dose of 600 mg, while 19 had dose reduction to 300 mg due to adverse events. Sixty-one participants were female, 71 were HIV-positive, and their median weight was 56 kg (interquartile range [IQR], 50 to 63). In the final model, typical values for clearance and central volume were 3.57 liters/h and 40.2 liters, respectively. HIV coinfection had no significant effect on linezolid exposure. Simulations showed that 600-mg dosing achieved the efficacy target (area under the concentration-time curve for the free, unbound fraction of the drug [[Formula: see text] at a MIC level of 0.5 mg/liter) with 96% probability but had 56% probability of exceeding safety target ([Formula: see text]. The 300-mg dose did not achieve adequate efficacy exposures. Our model characterized population pharmacokinetics of linezolid in South African patients with DR-TB and supports the 600-mg daily dose with safety monitoring.
Keywords: NONMEM; modeling and simulation; optimized dosing regimen; popPK/PD; population pharmacokinetics; tuberculosis.