Clinical performance of the Abbott RealTime Mycobacterium tuberculosis (MTB) PCR on bronchoscopic specimens for diagnosing pulmonary tuberculosis

Abstract Purpose We evaluated the performance of the Abbott RealTi me Mycobacterium tuberculosis (MTB) PCR (RT MTB) on bronchoscopic specimens using conventional culture as the reference standard in a low–Tuberculosis (TB) prevalence setting. Methods A total of 6,988 specimens (4,682 bronchial aspirates [BAS] and 2,306 bronchoalveolar lavages [BAL]) from 4,118 patients with suspected pulmonary TB were included. When BAS and BAL specimens from the same bronchoscopy procedure were available, these were mixed 1:1 prior to culture inoculation and PCR testing. Following processing, specimens were inoculated into a Löwenstein-Jensen and a Bactec MGIT 960 tube and incubated at 37 °C for 3 months and at 35 °C for 8 weeks, respectively. RT MTB was performed as indicated by the manufacturer. RT MTB targets the insertion sequence IS 6110 and the protein antigen B (PAB) gene, both highly conserved within the Mycobacterium tuberculosis complex. Whole genome Next generation sequencing of clinical MTBC isolates was performed when indicated. Results Among the 104 culture-positive specimens, 84 (1.2%) were detected by PCR. Additionally, 16 specimens (0.3% of all samples), from 16 patients, were PCR-positive despite negative culture results. Conversely, 20 specimens (0.4% of all samples), from 19 patients, were culture-positive but not detected by PCR. No significant differences were found between PCR-positive and PCR-negative specimens with respect to the number of IS 6110 copies per isolate or PAB gene sequences ( P = 0.69). Finally, there were 4,683 specimens (97.4%) from the remaining 4,014 patients tested PCR-negative/Culture-negative. Following the resolution of discrepancies based on clinical grounds the sensitivity and specificity of RT MTB were 83.9% (CI 95%, 76.0–90.0) and 99.9% (CI 95%, 99.9–99.9), respectively. These results exceed the minimum performance requirements defined in the WHO Target Product Profiles for molecular TB diagnostics. Although RT MTB is not a point-of-care test but rather a moderate-complexity automated NAAT, it is recommended by WHO as part of the Abbott RealTime MTB/MTB RIF-INH testing algorithm, in which MTBC detection by RT MTB is followed by reflex testing with the MTB RIF/INH assay for detection of rifampicin and isoniazid resistance. Conclusion RT MTB shows a good performance on bronchoscopic specimens.