ABSTRACT
OBJECTIVES: Cross-reactivity with nontuberculous mycobacteria (NTM) species might limit the use of urine lipoarabinomannan (LAM) test to diagnose tuberculosis (TB) in people living with HIV (PLWH). This study aimed to investigate the utility of the LAM test among hospitalized HIV-positive patients. METHODS: This prospective study enrolled HIV-positive inpatients with any TB symptom or seriously ill patients with advanced immunodeficiency. Urine samples were tested using the Alere Determine LAM Ag, and participants were categorized as confirmed TB, confirmed NTM infection, unclassified mycobacteria infection, and no mycobacteria infection based on microbiologic reference standards. RESULTS: A total of 382 participants were included. The prevalence of confirmed TB and NTM infection was 5.24% (20 of 382) and 4.45% (17 of 382), respectively. The sensitivity and specificity of the urine LAM for TB diagnosis were 65.00% (95% confidence interval [CI] 40.78-84.61) and 89.36% (95% CI 85.68-92.36), respectively. The LAM test for NTM yielded a sensitivity of 58.82% (95% CI 32.92-81.56) and specificity of 88.61% (95% CI 84.87-91.70). Notably, the negative predictive values of the urine LAM for TB and NTM were 97.85% (95% CI 95.63-99.13) and 97.85% (95% CI 95.63-99.13), respectively. CONCLUSIONS: Cross-reactivity with NTM cause high false-positive LAM for TB diagnosis in PLWH. The correct identification of mycobacteria species is crucial for deciding treatment strategies.