Sequencing of Pleural Fluid and Plasma for Tuberculous Pleuritis

Background The laboratory diagnosis of tuberculous pleuritis (TBP) is hindered by the paucibacillary nature of Mycobacterium tuberculosis in the pleural space, resulting in low sensitivity of microbiological culture and polymerase chain reaction-based analyses on pleural fluid. The use of metagenomic next-generation sequencing for diagnosing TBP may be limited by the background noise of DNA from nontuberculous mycobacteria. Methods We performed targeted sequencing to analyze M. tuberculosis DNA in paired pleural fluid and plasma from prospectively enrolled consecutive patients with new-onset pleural effusion. We used a bioinformatics alignment algorithm to the M. tuberculosis genome that was masked for regions with high sequence similarity to nontuberculous mycobacteria. Our primary outcome was a comparison of diagnostic sensitivity between M. tuberculosis sequencing as described above and culture using McNemar's test. Results Among the included 329 patients with pleural effusion, 34 patients with TBP were identified. Targeted sequencing detected M. tuberculosis DNA fragments in the pleural fluid of all TBP cases (median, 267.6 reads per 10 million [RP10M]; interquartile range [IQR], 30.8-2644.3) but absent in 288 out of 295 (97.6%) non-TBP samples (median, 0 RP10M; IQR, 0-0). Targeted sequencing of pleural fluid achieved a sensitivity of 97.1% for TBP detection at a cutoff of 2 RP10M, in contrast to 47.1% by M. tuberculosis culture (P Conclusions Targeted sequencing of pleural fluid with selectively masked M. tuberculosis genomic alignment accurately diagnosed TBP and outperformed conventional diagnostic tests. (Supported by InnoHK and the Hong Kong Tuberculosis, Chest and Heart Diseases Association; ClinicalTrials.gov number, NCT05397730.).