Association between acquired rifamycin resistance and the pharmacokinetics of rifabutin and isoniazid among patients with HIV and tuberculosis

Background. The occurrence of acquired rifamycin resistance despite use of directly observed therapy for tuberculosis is associated with advanced human immunodeficiency virus (HIV) disease and highly intermittent administration of antituberculosis drugs. Beyond these associations, the pathogenesis of acquired rifamycin resistance is unknown. Methods. We performed a pharmacokinetic substudy of patients in a trial of treatment with twice-weekly rifabutin and isoniazid. Results. A total of 102 (60%) of 169 patients in the treatment trial participated in the pharmacokinetic substudy, including 7 of 8 patients in whom tuberculosis treatment failure or relapse occurred in association with acquired rifamycin-resistant mycobacteria (hereafter, "ARR failure or relapse"). The median rifabutin area under the concentration-time curve (AUC(0-24)) was lower for patients with than for patients without ARR failure or relapse (3.3 vs. 5.2 mu g*h/mL; P = .06, by the Mann-Whitney exact test). In a multivariate analysis adjusted for CD4(+) T cell count, the mean rifabutin AUC(0-24) was significantly lower for patients with ARR failure or relapse than for other patients (3.0 mu g*h/mL [95% confidence interval {CI}, 1.9-4.5] vs. 5.2 mu g*h/mL [95% CI, 4.6-5.8]; P = .02, by analysis of covariance). The median isoniazid AUC(0-12) was not significantly associated with ARR failure or relapse (20.6 vs. 28.0 mu g*h/mL; P = .24, by the Mann-Whitney exact test). However, in a multivariate logistic regression model that adjusted for the rifabutin AUC(0-24), a lower isoniazid AUC(0-12) was associated with ARR failure or relapse (OR, 10.5; 95% CI, 1.1-100; P = .04). Conclusions. Lower plasma concentrations of rifabutin and, perhaps, isoniazid were associated with ARR failure or relapse in patients with tuberculosis and HIV infection treated with twice-weekly therapy.