Limited value of whole blood Xpert(®) MTB/RIF for diagnosing tuberculosis in children

Abstract

Objectives: We evaluated the ability of the Xpert(®) MTB/RIF assay to detect Mycobacterium tuberculosis in whole blood of children with tuberculosis in tuberculosis endemic settings with high rates of HIV infection.

Methods: From June 2011 to September 2012 we prospectively enrolled children with symptoms or signs suggestive of tuberculosis at three research centres in Tanzania and Uganda. After clinical assessment, respiratory specimens were collected for microscopy and culture, as well as whole blood for Xpert(®) MTB/RIF. Children were classified according to standardised case definitions.

Results: A total of 232 children were evaluated; 14 (6.0%) had culture-confirmed tuberculosis. The Xpert(®) MTB/RIF assay detected M. tuberculosis in 5/232 (2.2%) blood samples with 1 (0.4%) error reading and presumably 1 (0.4%) false-positive result. The sensitivity of the assay in children with culture-confirmed (1/14) versus no tuberculosis (1/117) was 7.1% (95% CI, 1.3-31.5). Three of the five Xpert(®) MTB/RIF positive patients had negative cultures, but were classified as probable tuberculosis cases. Assay sensitivity against a composite reference standard (culture-confirmed, highly probable or probable tuberculosis) was 5.4% (95% CI, 2.1-13.1).

Conclusion: Whole blood Xpert(®) MTB/RIF demonstrated very poor sensitivity, although it may enhance the diagnostic yield in select cases, with culture-negative tuberculosis.

Keywords: Childhood; Diagnosis; Tuberculosis; Whole blood; Xpert MTB/RIF.