Clinical Application of Genexpert on BAL Samples in Management of TB in Intermediate Burden Area

Abstract

SESSION TITLE: Tuberculosis SESSION TYPE: Original Investigation Poster PRESENTED ON: Wednesday, November 1, 2017 at 01:30 PM - 02:30 PM PURPOSE: Assess the GeneXpert from clinical perspective METHODS: Limited experience presently exists on clinical use of GeneXpert (Xpert) on bronchoalveolar lavage (BAL) fluid samples obtained from patients clinically suspected of pulmonary TB. We recruited sputum acid-fast-bacilli (AFB) smear negative patients who were offered bronchoscopy. BAL fluid was collected for AFB smear, TB culture and conventional TB PCR. An extra 5 ml BAL fluid was collected for Xpert. Performance of Xpert was compared to each of the other methods of detection. Cases were considered to have TB if any one of the above samples was positive (overall samples). RESULTS: From October 2015 to October 2016, a total of 211 samples with complete data for analysis were collected. Patients’ mean+SD age was 60.4+14.9 years, of which 136 were males and 75 were females. Cough and haemoptysis were the presenting symptoms in 70% and 37.4% respectively. Apical shadows in chest X-ray (CXR) were more common in Xpert positive cases (X2=10, p=0.02). For CT abnormalities, there were no significant differences in presence/absence of ground glass, tree in buds and consolidation, between Xpert positive or negative cases. However, there were more cavitations and pleural effusions in Xpert positive cases (X2=10 and 12, p=0.02 and 0.05 respectively). Sensitivities of Xpert ranged from 83 to 100% when compared to each other category of testing, with 79% for overall samples. Similar comparisons showed specificities ranged from 85.3% to 96%, while overall specificity was 98%. The positive predictive values of Xpert compared to each other category of testing ranged from 18.9% to 81.1%, with overall value of 91.9%, while negative predictive values ranged from 96.5% to 100%, with overall value of 94.8%. There was no rifampicin resistance identified. CONCLUSIONS: From practical point of view in clinical practice, treatment would be started with any positive result. It would therefore be more useful to focus on overall performance rather than individual comparison values. Out of the 9 false negative Xpert samples (8 were conventional TB PCR negative): 6 were diagnosed by BAL culture, 1 by biopsy which showed necrotizing granuloma, 1 by AFB positive pleural biopsy, and another one by conventional TB PCR. These cases helped explain the relatively lower sensitivities in specific situations. On the other hand, comparison of performance with conventional TB PCR showed Xpert’s relative higher specificity. There were only 3 false positive cases, in which the MTB detection was low, probably cases with low bacillary load. Of these, 2 were put on empirical treatment with favorable clinical responses, while one defaulted follow up. If the clinical responses were also taken as evaluation, then the positive predictive values would be even higher. CLINICAL IMPLICATIONS: In summary, in an intermediate TB burden area like Hong Kong, Xpert has high specificity and similar performance as conventional PCR. Technical simplicity made it useful tool to rule in TB for initiating early treatment in clinical suspicious cases. However, it cannot replace other investigations as the only test for diagnosing pulmonary TB. DISCLOSURE: The following authors have nothing to disclose: Kin Wang To, Kai Man Kam, shui shan Lee, Ka Pang Chan, Tommy Yip, Rachel Lo, Jenny Ngai, Susanna Ng No Product/Research Disclosure Information