Purpose: To determine the diagnostic accuracy of the Xpert MTB/RIF assay in patients with smear-negative pulmonary tuberculosis (TB) and to assess clinical and CT characteristics of Xpert-negative pulmonary TB.
Material and methods: We retrospectively reviewed the records of 1,400 patients with suspected pulmonary TB for whom the sputum Xpert MTB/RIF assay was performed between September 1, 2014 and February 28, 2020. Clinical and CT characteristics of smear-negative pulmonary TB patients with negative Xpert MTB/RIF results were compared with positive results.
Results: Of 1,400 patients, 365 (26.1%) were diagnosed with pulmonary TB and 190 of 365 patients (52.1%) were negative for sputum acid-fast bacilli. The diagnosis of pulmonary TB was based on a positive culture, positive Xpert MTB/RIF or the clinical diagnoses of patients treated with an anti-TB medication. The sensitivity, specificity, positive predictive and negative predictive values of sputum Xpert MTB/RIF for smear-negative pulmonary TB were 41.1%, 100%, 100%, and 90.1%, respectively. Finally, 172 patients with smear-negative pulmonary TB who underwent chest CT within 2 weeks of diagnosis were included to compare Xpert-positive (n = 66) and Xpert- negative (n = 106) groups. Patients with sputum Xpert-negative TB showed lower positive rates for sputum culture (33.0% vs. 81.8%, p<0.001) and bronchoalveolar lavage culture (53.3% vs. 84.6%, p = 0.042) than in Xpert-positive TB. Time to start TB medication was longer in patients with Xpert-negative TB than in Xpert-positive TB (11.3±16.4 days vs. 5.0±8.7 days, p = 0.001). On chest CT, sputum Xpert-negative TB showed significantly lower frequency of consolidation (21.7% vs. 39.4%, p = 0.012), cavitation (23.6% vs. 37.9%, p = 0.045), more frequent peripheral location (50.9% vs. 21.2 p = 0.001) with lower area of involvement (4.3±4.3 vs. 7.6±6.4, p<0.001). Multivariate analysis revealed peripheral location (odds ratios, 2.565; 95% confidence interval: 1.157-5.687; p = 0.020) and higher total extent of the involved lobe (odds ratios, 0.928; 95% confidence interval: 0.865-0.995; p = 0.037) were significant factors associated with Xpert MTB/RIF-negative TB. Regardless of Xpert positivity, more than 80% of all cases were diagnosed of TB on chest CT by radiologists.
Conclusion: The detection rate of sputum Xpert MTB/RIF assay was relatively low for smear negative pulmonary TB. Chest CT image interpretation may play an important role in early diagnosis and treatment of Xpert MTB/RIF-negative pulmonary TB.