Protease Inhibitors and Rifabutin: Isn't the Jury Still Out?

Use of rifabutin with protease inhibitors for human immu-nodeficiency virus-infected patients with tuberculosis. Sir—The attempt by Narita et al. [1] to delineate the pharmacokinetic interactions between HIV type 1 protease in-hibitors (PIs) and rifabutin administered intermittently addresses the practical question of juggling intermittent directly observed therapy (DOT) for tuberculosis (TB) with PI-based highly active antire-troviral therapy (HAART). Although the authors provide data that reassure that toxicity was not problematic and that vi-rologic outcomes were reasonably successful in a supervised inpatient facility, their data do not provide unequivocal support for the use of PIs with rifabutin. The authors measured 2-h levels of PI, rather than trough levels. Since rifabutin speeds up the hepatic metabolism of PIs more than their intestinal metabolism [2], PI trough levels are preferentially decreased. Virologic failure with the use of PIs is thought to be related to the extent to which PI trough levels fall short of the inhibitory concentration of the drug for the virus [3]. Without demonstration of adequate PI trough levels, the adequacy of PI levels with rifabutin is still in doubt. Furthermore, the authors' data showed equivalence in the 2-h levels of indinavir, 1200 mg q8h, which is the dosage they recommend with rifabutin, and indina-vir, 1200 mg q12h, which is a dosage that has been discouraged by indinavir's manufacturer since that dosage was associated with an unacceptable rate of virologic failure (Merck, data on file). The authors showed virologic suppression to the level of 500 copies/mL. Since durability of viral suppression by antire-troviral therapy relates to a nadir virus load [4], the assay that uses 500 copies/ mL cannot be considered the gold standard by which to validate a regimen. Of concern was that only 4 of 8 antiretroviral therapy (ART)–naı¨ve patients achieved an undetectable virus load while receiving a PI and rifabutin. The patients who were not ART-naı¨ve fared better, but, as a group, they had a lower virus load in comparison with the naı¨ve patients. Our experience at Jacobi Medical Center (Bronx, NY) illustrates practical obstacles to delivering HAART to patients with TB and loss of virologic suppression in a small number of patients concurrently treated with rifabutin and a PI. We have cared for 32 patients with HIV/TB infection since 1996 [5]. Fifteen patients had known HIV infection prior to TB; only 5 of these patients were receiving HAART. Seventeen had newly diagnosed HIV infection, consequent to TB. Seven …