PurposeCartridge based nucleic acid amplification test (CBNAAT) has been endorsed by the WHO as the screening test for diagnosing extrapulmonary tuberculosis (EPTB). In the present study we report the agreement between CBNAAT (Xpert MTB/RIF), liquid culture (LC) and line probe assay (LPA) for diagnosis of Mycobacterium tuberculosis and detection of drug resistance among EPTB cases. MethodsThe EP samples were subjected to CBNAAT (Xpert MTB/RIF, Cepheid, USA) and wherever possible, to LC (MGIT 960, Becton Dickinson, USA) followed sequentially by first line and second line-LPA (FL-LPA, SL-LPA, Hain Lifescience, Germany) on the isolates. ResultsTotal 566/4080 (13.9%) EP samples were detected positive for M. tuberculosis on CBNAAT. Aspirates from lymph nodes were most often positive (11/30; 36.6%), followed by pus (240/873; 27.5%) and CSF samples (166/104; 15.8%). The detection of M. tuberculosis was more in adults than children except in tissue biopsy samples. Rifampicin resistance was also higher among adults except CSF in which resistance was more in children. Total 185 of 566 (32.7%) CBNAAT positive and 770 of 3510 (21.9%) CBNAAT negative samples could be cultured of which 110/185 (59.4%) and 33/770 (4.3%) respectively turned positive. FL-LPA and SL-LPA of 143 culture isolates showed that 27 isolates had drug resistance, of which 3 (2.1%) were XDR, 11 (7.7%) were Pre-XDR (FQ) and 13 (9.1%) were MDR. Of these 27 resistant isolates, 12 were negative by CBNAAT and two were mislabeled as Rifampicin sensitive or indeterminate based on the unique RpoB gene mutation patterns on LPA. The positive and negative agreements between LC and CBNAAT for detection of M. tuberculosis were 67.1% and 92.7% respectively and between LPA and CBNAAT for rifampicin resistance detection were 98.9% and 92.9% respectively. ConclusionsFor EPTB, CBNAAT should be accompanied with LC wherever possible irrespective of the CBNAAT result.