Predictors and outcomes of Mycobacterium tuberculosis bacteremia among patients with HIV and tuberculosis co-infection enrolled in the ACTG A5221 STRIDE study

Background: We evaluated predictors and outcomes of Mycobacterium tuberculosis bacteremia among participants undergoing baseline mycobacterial blood culture in the ACTG A5221 STRIDE study, a randomized clinical trial comparing earlier with later ART among HIV-infected patients suspected of having tuberculosis with CD4-positive T-lymphocyte counts (CD4 counts) <250 cells/mm(3). We conducted a secondary analysis comparing participants with respect to presence or absence of M. tuberculosis bacteremia. Methods: Participants with a baseline mycobacterial blood culture were compared with respect to the presence or absence of M. tuberculosis bacteremia. Baseline predictors of M. tuberculosis bacteremia were identified and participant outcomes were compared by mycobacteremia status. Results: Of 90 participants with baseline mycobacterial blood cultures, 29 (32.2%) were female, the median (IQR) age was 37 (31-45) years, CD4 count was 81 (33-131) cells/mm(3), HIV-1 RNA level was 5.39 (4.96-5.83) log(10) copies/mL, and 18 (20.0%) had blood cultures positive for M. tuberculosis. In multivariable analysis, lower CD4 count (OR 0.85 per 10-cell increase, p = 0.012), hemoglobin <= 8.5 g/dL (OR 5.8, p = 0.049), and confirmed tuberculosis (OR 17.4, p = 0.001) were associated with M. tuberculosis bacteremia. There were no significant differences in survival and AIDS-free survival, occurrence of tuberculosis immune reconstitution inflammatory syndrome (IRIS), or treatment interruption or discontinuation by M. tuberculosis bacteremia status. IRIS did not differ significantly between groups despite trends toward more virologic suppression and greater CD4 count increases at week 48 in the bacteremic group. Conclusions: Among HIV-infected tuberculosis suspects, lower CD4 count, hemoglobin <= 8.5 g/dL, and the presence of microbiologically confirmed pulmonary tuberculosis were associated with increased adjusted odds of mycobacteremia. No evidence of an association between M. tuberculosis bacteremia and the increased risk of IRIS was detected.