ABSTRACT
OBJECTIVE: To assess Xpert® MTB/RIF (Xpert) and Xpert® MTB/RIF Ultra (Ultra) performance in diagnosing pediatric tuberculous meningitis (TBM).METHODS: We conducted a study among children with suspected meningoencephalitis in Pune, India. Clinical, radiological, laboratory, and treatment data were analyzed to classify disease as definite, probable, possible or no TBM, using microbiologic or composite reference standards. We tested cerebrospinal fluid (CSF) either using Xpert or Ultra and estimated test performance characteristics.RESULTS: Of 341 participants, 149 (43.7%) were tested using Ultra and 192 (56.3%) with Xpert. Ultra had higher sensitivity (50% vs. 18%), lower specificity (91% vs. 99%), poor positive predictive value (PPV) (13% vs. 75%), and higher negative predictive value (NPV) (99% vs. 93%) than Xpert using the composite reference standard, with similar results by the microbiologic reference standard. Of 10 participants with trace positivity on Ultra, none met clinical TBM definitions.CONCLUSION: This is the first study to report on diagnostic performance of Ultra in pediatric TBM, which showed higher sensitivity and NPV than Xpert. For children presenting with nonspecific clinical features, Ultra is a promising diagnostic test. Further studies are required to define its optimal clinical use, including interpretation of trace positive results.
