Evaluation of two commercial nucleic acid amplification tests for diagnosis of extrapulmonary tuberculosis from lymph nodes of children in Tanzania

Abstract

Approximately one third of the global population is latently infected with tuberculosis (TB) and there are approximately 9.6 million new cases of TB disease per year resulting in 1.5 million deaths. Eleven percent of cases globally occur in children and 81% of the burden of TB disease is borne by the developing world and countries with emerging economies (BRICS). The African region accounts for 28% of new TB cases globally. TB remains a significant public health concern globally, particularly amongst children and immunocompromised individuals. Diagnosis of childhood TB is an on going challenge, as children usually do not present with the same signs and symptoms as adults, and are often misdiagnosed. Tuberculosis infection in children is seldom confirmed through sputum culture, as good sputum samples can rarely be collected. Only 15% of cases from children are sputum smear positive by acid-fast staining, and only 30%–40% are Mycobacterium tuberculosis (MTB) culture positive. Up to 40% of children present with extrapulmonary manifestations of TB disease. The most common manifestation is tuberculous lymphadenopathy. Good specimens for TB detection can be obtained from these cases through fine needle aspiration biopsy (FNAB), a cost-effective and practical out patient procedure for obtaining specimens from enlarged superficial lymph nodes. The conventional laboratory techniques that have been used globally for TB diagnosis are Ziehl-Neelsen (ZN) staining for acid fast bacilli (AFB) microscopy, culture and more commonly now LED microscopy, cytological determination with auto-fluorescence staining and molecular Xpert® MTB/RIF (Cepheid) detection. AFB-microscopy requires minimally 5000 AFB per millilitre of specimen to yield a consistently positive result and observation of between 100 and 300 microscopic fields in order to obtain accurate results. Culture for MTB is the gold standard diagnostic method but has a slow turnaround time and requires laboratory resources that are not available in most parts of the world. Recent systematic reviews of studies evaluating commercially available nucleic acid amplification test (NAAT) technologies confirm very high specificity, with sensitivity approaching, but not reaching, that of culture. The complexity and insufficient robustness of existing commercial NAAT protocols and their need for precision instruments, a high degree of technical support, and quality assurance make them unsuitable for most low resource TB endemic countries. In addition, none of these techniques have been fully validated for diagnosing TB in children and specifically not for extrapulmonary specimens. In light of these challenges, there is promise in two technologies that have been developed and under evaluation over the last few years; the Xpert® MTB/RIF kit (Cepheid) for rapid detection of MTB and rifampicin resistance endorsed for use by the WHO; and the Ustar EasyNATTM TB IAD kit (Ustar Biotechnologies, Hangzhou) for detection of MTB, selected by the WHO as a technology on assessment for use in TB endemic countries. In this study, their performance was assessed against the conventional laboratory diagnostic techniques of smear AFB microscopy, cytology and mycobacterial culture of fine needle aspirates from lymph nodes of children suspected of TB lymphadenopathy in Tanzania. Age defined clinical assessments were done for all 75 participants and TB treatment initiated based on these and/or laboratory diagnostic outputs. All laboratory diagnostic modalities were primarily assessed against TB culture as the conventional reference standard. As has been evidenced in earlier studies, the sensitivities for both smear microscopy and TB culture were very low in these extrapulmonary specimens. Lacking a true reference standard, composite reference standards (CRSs) were created to assess the performance of the test modalities under study. An alternative method for assessing diagnostic accuracy under these conditions is latent class analyses (LCA), which was utilized to further assess the performance of all diagnostic modalities in the study. The overall outcomes of the project demonstrated that cytomorphology was a feasible and effective technique for detection of TB in lymph node aspirates (sensitivity: LCA 100%, specificity: LCA 94.7%) that may complement TB culture (sensitivity: LCA 74.5%, specificity: LCA 90.3%). Further, it was shown that Xpert (sensitivity: CRS 58% LCA 70.7%, specificity: CRS 93% LCA 94.2%) was superior in performance to EasyNAT (sensitivity: CRS 19% LCA 29.2%, specificity: CRS 100% LCA 100%) and ZN (sensitivity: CRS 14% LCA 19.1%, specificity: CRS 100% LCA 100%) analyses, respectively. Combining two or more tests significantly improved the diagnostic efficacy, but including either EasyNAT testing or ZN microscopy to a diagnostic algorithm that already had Xpert testing added no value. These findings indicate that combining clinical assessment, cytology and Xpert MTB/RIF can provide for rapid and accurate diagnosis of childhood tuberculous lymphadenitis. Larger diagnostic evaluation studies on Xpert MTB/RIF would be required, to assess its use as a solitary initial test for tuberculous lymphadenitis in children.