Background: Case-control studies have noted an association between HIV and acquired rifamycin resistance (ARR) in TB patients (pts); ARR has also been noted in HIV/TB pts treated with intermittent INH plus rifabutin or rifapentine in the continuation phase of therapy. However, there has not been a comparison of the risk of ARR with rifampin vs rifabutin, the 2 rifamycins most commonly used to treat TB in HIV⁺ persons.

Methods: We conducted a retrospective cohort study of all persons with culture + rifamycin-susceptible TB who completed a course of directly-observed therapy in Baltimore from 1/1/1993–12/31/2001. Treatment was daily for 3 weeks, then twice weekly. Recurrent TB was defined as developing a + M. tuberculosis culture after converting to culture-negative during the initial course of treatment. ARR was defined as developing rifamycin resistance after having had rifamycin-susceptible disease. DNA fingerprinting was performed 1994-present. Categorical variables were compared with c² or Fisher’s exact test; continuous variables with the Mann-Whitney U test.

Results: During the study period, 618 culture + TB cases occurred; 431 met the inclusion criteria: 109 (25%) HIV⁺, 182 (42%) HIV^–, and 140 (33%) HIV-unknown. Demographic and clinical factors of HIV-negatives and HIV-unknowns were similar, so these 2 groups were combined. There were 16/431 (3.7%) TB recurrences; 9/109 (8.3%) HIV positives vs 7/322 (2.2%) HIV-negative/unknown (RR = 2.3; p = 0.007). DNA fingerprinting was available for 9/16 pts; the 1^st and 2^nd isolates matched in all 9. Among HIV positives, relapse was only associated with low median initial CD4 count (51/mm³ in relapsers vs 137/mm³ in non-relapsers; p = 0.02) and not with the rifamycin used. Among HIV-negative/unknowns, sputum culture + after 2 months of treatment, cavitary pulmonary disease, and white race were associated with relapse. 3/109 (2.8%) HIV-positives had ARR (median CD4: 61/mm³), compared to 0/322 HIV-negative/unknown (RR = 4.0; p = 0.02). Of the 81 HIV-positives who received rifampin, 3 (3.7%) had ARR, compared to 0/27 pts who received rifabutin (p = 0.57).

Conclusion: Among HIV⁺ TB pts, low CD4 count was the only risk factor for TB relapse; the risk factors in HIV-negative/unknowns were the same as those previously reported. ARR was seen only in HIV⁺ pts with low CD4 counts, and was not higher with rifabutin-based than rifampin-based regimens. Further studies are needed to identify how to prevent ARR.