Abstract
BACKGROUND: The WHO recently recommended the new Xpert MTB/RIF Ultra assay (Ultra) instead of the Xpert MTB/RIF assay because Ultra has improved sensitivity. We report the diagnostic accuracy of Ultra for tuberculous adenitis in a tuberculosis and HIV endemic setting. METHODS: We obtained fine-needle aspirates (FNA) and lymph node tissue by core-needle biopsy in adult patients with peripheral lymphadenopathy of >20 mm. Ultra and mycobacterial culture were performed on FNA and tissue specimens, with histological examination of tissue specimens. We assessed the diagnostic accuracy of Ultra against a composite reference standard of 'definite tuberculosis' (microbiological criteria) or 'probable tuberculosis' (histological and clinical criteria). RESULTS: We prospectively evaluated 99 participants of whom 50 were HIV positive: 21 had 'definite tuberculosis', 15 'probable tuberculosis' and 63 did not have tuberculosis (of whom 38% had lymphoma and 19% disseminated malignancy). Using the composite reference standard the Ultra sensitivity on FNA was 70% (95% CI 51-85; 21 of 30), and on tissue was 67% (45-84; 16/24) these were far superior to the detection of acid-fast bacilli on an FNA (26%; 7/27); AFB on tissue (33%; 8/24); or tissue culture (39%; 9/23). The detection of granulomas on histology had high senstivity (83%) but the lowest specficity. When compared with culture the Ultra on FNA had a sensitvity of 78% (40-97; 7/9) and tissue 90% (55-100; 9/10). CONCLUSIONS: Ultra performed on FNA or tissue of a lymph node had good sensitivity and high specificity. Ultra had a higher yield than culture and has the advantage of being a rapid test. Ultra on FNA would be an appropriate initial investigation for lymphadenopathy in tuberculosis endemic areas followed by a core biopsy for histopathology with a repeat Ultra on tissue if granulomas are present.