Background: Tuberculosis (TB) remains the leading cause of death among HIV-infected children in sub-Saharan Africa. Accurate and prompt identification of TB infection remains challenging, particularly among young children who are often unable to produce sputum. In this study, we evaluated the sensitivity and specificity of the lipoarabinomannan (LAM) TB ELISA for the detection of mycobacteria-specific antigen in urine samples of HIV-infected Nigerian children.
Methodology: As part of routine care, all HIV-infected children enrolling in the APIN/Harvard PEPFAR program undergo routine screening for TB consisting of: clinical evaluation for symptoms or signs of TB infection, chest radiograph, and tuberculin skin testing (TST). In addition, for those children with a chronic productive cough, sputum samples are collected when possible. A composite of these standard TB diagnostic techniques was used as the “gold standard“, and compared against results of the urine LAM assay.
Results: 184 HIV-infected pediatric patients were enrolled at two study sites; 40% were female with a median age of 5 years and median baseline CD4+cell count of 434 cells/mm3. Almost 36% of patients (66 children) were diagnosed with TB within 3 months after collecting the LAM sample. The sensitivity of the LAM assay to detect the composite diagnosis of TB infection is 10.6% (95% CI 4.7-21.2%), specificity is 97.4% (95% CI 91.9-99.3%), and the PPV is 70% (95% CI 35.3-91.9%). The median CD4 count was 103 cells/mm3 (IQR, 25-331 cells/mm3) among children with a positive LAM, compared to 393 cells/mm3 (IQR, 177-717 cells/mm3) among those diagnosed with TB but with a negative LAM result.
Conclusions: Although the sensitivity of the LAM assay was low in this population, the relatively high PPV suggests that a positive result may add valuable information to the care of a sick child. However, more sensitive rapid diagnostic methods for the detection of TB in young children are urgently needed.