TB testing in HIV-positive patients prior to antiretroviral treatment.

TB diagnosis in patients with HIV is challenging due to the lower sensitivities across tests. Molecular tests are preferred and the Xpert MTB/RIF assay has limitations in lower-income settings. We evaluated the performance of loop-mediated isothermal amplification (LAMP) and the lipoarabinomannan (LAM) test in HIV-positive, ART-naïve clinic patients. A total of 783 eligible patients were enrolled; three spot sputum samples of 646 patients were tested using TB-LAMP, Xpert, smear microscopy and culture, while 649 patients had TB-LAM testing. Sensitivity, specificity, and negative and positive predictive values were estimated with 95% confidence intervals. Sensitivities for smear microscopy, TB-LAMP and Xpert were respectively 50%, 63% and 74% compared to culture, with specificities of respectively 99.2%, 98.5% and 97.5%. An additional eight were positive on TB-LAM alone. Seventy TB patients (9%) were detected using standard-of-care testing, an additional 27 (3%) were detected using study testing. Treatment was initiated in 57/70 (81%) clinic patients, but only in 56% (57/97) of all those with positive TB tests; 4/8 multidrug-resistant samples were detected using Xpert. TB diagnostics continue to miss cases in this high-burden setting. TB-LAMP was more sensitive than smear microscopy, and if followed by culture and drug susceptibility testing as required, can diagnose TB in HIV-positive patients. TB-LAM is a useful add-in test and both tests at the point-of-care would maximise yield.