RESUMEN
In adults requiring protease inhibitor (PI)-based antiretroviral therapy (ART), replacing rifampicin with rifabutin is a preferred option, but there is lack of evidence to guide rifabutin dosing in children, especially with PIs. We aimed to characterize the population pharmacokinetics of rifabutin and 25-O-desacetyl rifabutin (des-rifabutin) in children and optimize its dose. We included children from three age cohorts: (i) <1-year-old cohort and (ii) 1- to 3-year-old cohort, who were ART naïve and received 15- to 20-mg/kg/day rifabutin for 2 weeks followed by lopinavir/ritonavir (LPV/r)-based ART with 5.0- or 2.5 mg/kg/day rifabutin, respectively, while the (iii) >3-year-old cohort was ART-experienced and received 2.5-mg/kg/day rifabutin with LPV/r-based ART. Non-linear mixed-effects modeling was used to interpret the data. Monte Carlo simulations were performed to evaluate the study doses and optimize dosing using harmonized weight bands. Twenty-eight children were included, with a median age of 10 (range 0.67-15.0) years, a median weight of 11 (range 4.5-45) kg, and a median weight-for-age z score of -3.33 (range -5.15 to -1.32). A two-compartment disposition model, scaled allometrically by weight, was developed for rifabutin and des-rifabutin. LPV/r increased rifabutin bioavailability by 158% (95% confidence interval: 93.2%-246.0%) and reduced des-rifabutin clearance by 76.6% (74.4%-78.3%). Severely underweight children showed 26% (17.9%-33.7%) lower bioavailability. Compared to adult exposures, simulations resulted in higher median steady-state rifabutin and des-rifabutin exposures in 6-20 kg during tuberculosis-only treatment with 20 mg/kg/day. During LPV/r co-treatment, the 2.5-mg/kg/day dose achieved similar exposures to adults, while the 5-mg/kg/day dose resulted in higher exposures in children >7 kg. All study doses maintained a median Cmax of <900 µg/L. The suggested weight-band dosing matches adult exposures consistently across weights and simplifies dosing.