Diagnostic Yield of urine Xpert MTB/RIF ultra to detect Mycobacterium tuberculosis among severely immunosuppressed inpatients with HIV: A prospective cohort study in Ghana

Objectives Due to the higher prevalence of sputum scarce pulmonary tuberculosis (TB) and extrapulmonary TB among people with HIV compared to HIV-negative people, diagnosis of HIV-associated TB cannot rely solely on sputum-based methods. We estimated the TB diagnostic yield of urine Xpert MTB/RIF Ultra (Ultra) and its prognostic significance among inpatients with HIV. Methods Adults admitted with HIV in Ghana between October 2019 and November 2021 were enrolled if they had a positive WHO 4-symptom screen, advanced HIV, or severe illness according to WHO criteria for people with HIV. Urine samples were frozen and later analyzed using Ultra. Two TB reference standards were used: an extended microbiological reference (eMRS) including culture, Xpert/Ultra on non-urine samples, or urine LF-LAM; and a composite reference (CRS) including eMRS or TB diagnosis by microscopy, recommended TB treatment, or TB at death. Patients were followed for 8 weeks. Results Of 141 participants, 118 (83.7%) provided urine samples for storage. TB prevalence was 25.0% by eMRS and 37.3% by CRS. Urine Ultra was positive in 34/118 (28.8%) patients. Ultra-positive patients had lower median CD4 count compared to Ultra-negative patients (31 vs 78 cells/mm³, P = 0.025). If urine Ultra had been used to guide TB diagnosis, it would have identified an additional 20 patients not detected by CRS. No difference in 8-week mortality was observed between Ultra-positive and negative patients. Conclusion Urine Ultra shows promise as a supplementary point-of-care TB test among inpatients with HIV.