Acquired rifampicin resistance in thrice-weekly antituberculosis therapy: impact of HIV and antiretroviral therapy.

Abstract

Risk factors for acquired rifampicin resistance (ARR) in human immunodeficiency virus (HIV)/tuberculosis coinfection, in the highly active antiretroviral therapy (HAART) era, needs evaluation. We studied the impact of HIV and HAART on ARR among patients taking thrice-weekly antituberculosis therapy. This cross-protocol analysis included patients with newly diagnosed, rifampicin-susceptible pulmonary tuberculosis, with and without HIV, enrolled in clinical trials (who took >80% of medication) at the National Institute for Research in Tuberculosis between 1999 and 2013. All patients received rifampicin and isoniazid for 6 months reinforced with pyrazinamide and ethambutol in the first 2 months, given thrice-weekly throughout the study along with HAART in one of the groups. Outcomes were categorized and multivariate logistic regression analysis performed to identify risk factors for ARR. The per-protocol results included patients with tuberculosis: 246 HIV-uninfected patients (HIV(-)TB(+)), 212 HIV patients not on HAART (non-HAART), and 116 HIV-infected patients on HAART. Median CD4 counts of the latter 2 groups were 150 and 93 cells/µL, respectively, and the median viral loads were 147 000 and 266 000 copies/mL, respectively. Compared with HIV(-)TB(+), the relative risks (RRs) for an unfavorable response in the coinfected, non-HAART and HAART groups were 2.1 (95% confidence interval [CI], 1.7-14.8; P<.0001) and 2.1 (95% CI, .9-5.2; P=.3), whereas for ARR, the RRs were 21.1 (95% CI, 2.6-184; P<.001) and 8.2 (95% CI, .6-104; P=.07), respectively. HIV-infected patients with tuberculosis treated with a thrice-weekly antituberculosis regimen are at a higher risk of ARR, compared with HIV-uninfected patients, in the presence of baseline isoniazid resistance. HAART reduces but does not eliminate the risk of ARR.