Abstract
BACKGROUND: Accurate non-sputum-based tuberculosis (TB) diagnostics are urgently needed to improve diagnostic yield and patient outcomes. OBJECTIVES: To compare the diagnostic accuracy and diagnostic yield of Urine Xpert Ultra (Urine-XPU) and Urine Determine(TM) TB Lipoarabinomannan (LAM) antigen test (AlereLAM) against both a microbiological and composite reference standard (MRS and CRS) in a rural, routine care setting in South Africa. METHOD: Adults (≥ 18 years) with HIV had sputum, urine and blood collected for comprehensive TB testing shortly after admission. Additionally, focused assessment with sonography for HIV-associated TB (FASH) was performed. The MRS was defined by Xpert Ultra or culture-based tests for Mycobacterium tuberculosis. The CRS incorporated these mycobacterial tests, FASH findings, and clinical response to empiric TB treatment. Follow-up was conducted at 3 months. RESULTS: A total of 206 participants were enrolled, with a median age of 39 years and 63% were female. Using the MRS the sensitivity of AlereLAM was 45.2% (95% confidence interval [CI]: 31.2-60.1) and Urine-XPU, 59.5% (95%CI: 44.5-73.0); and the specificity of AlereLAM was 93.6% (95%CI: 88.2-96.6) and Urine-XPU 95.0% (95%CI: 90.0% - 97.6%). Urine-XPU and AlereLAM performed better than sputum Xpert Ultra (Sputum-XPU) in patients with more severe illness. Additionally, Urine-XPU showed potential for accurately detecting rifampicin resistance. CONCLUSION: Urine-XPU and AlereLAM demonstrated comparable diagnostic accuracy for TB in hospitalised adults with HIV. Integrating Urine-XPU alongside AlereLAM and Sputum-XPU may improve timely and accurate diagnosis of TB and rifampicin resistance. Further research is required to optimise the diagnosis-to-treatment pathway.