Metagenomic next-generation sequencing: A promising tool for diagnosis and treatment of suspected pneumonia in rheumatic patients with acute respiratory failure: Retrospective cohort study

BACKGROUND: The effectiveness of metagenomic next-generation sequencing (mNGS) in respiratory pathogen detection and clinical decision-making in critically rheumatic patients remains largely unexplored. METHODS: A single-center retrospective study of 58 rheumatic patients who were admitted to ICU due to suspected pneumonia with acute respiratory failure if they underwent both bronchoalveolar lavage fluid specimen mNGS and combined microbiological tests (CMTs) was conducted to compare their diagnostic performance, using clinical composite diagnosis as the gold standard. Treatment modifications based on mNGS results were also reviewed. RESULTS: Forty-three patients were diagnosed with microbiologically confirmed pneumonia and 15 were considered as a non-infectious disease. mNGS outperformed CMTs in the accurate diagnosis of infectious and non-infectious lung infiltration (98.1% [57/58] vs. 87.9% [51/58], P = 0.031). A total of 94 causative pathogens were defined by the gold standard and 27 patients had polymicrobial pneumonia. The sensitivity of pathogen detection and complete concordance with the gold standard by mNGS exceeded those by CMTs (92.6% [87/94] vs. 76.6% [72/94], P < 0.001 and 72.1% [31/43] vs. 51.2% [22/43], P = 0.004, respectively). Moreover, 22 pathogens were detected only by mNGS and confirmed by orthogonal test. Accordingly, the etiological diagnosis changed in 19 cases, and the empirical treatment improved in 14 cases, including 8 cases of rescue treatment and 11 of antibiotics de-escalation. At the pathogen-type level, both methods were comparable for bacteria, but mNGS was advantageous to identify viruses (accuracy: 100% vs. 81%, P = 0.004). For Pneumocystis jirovecii detection, mNGS improved the sensitivity compared with Gomori’s methenamine silver stain (91.7% vs. 4.2%, P < 0.001) and was higher than polymerase chain reaction (79.2%), but the difference was not significant (P = 0.289). In terms of Aspergillus, the better sensitivity with a combination of culture and galactomannan test than that with mNGS was found (100% vs. 66.7%, P = 0.033). CONCLUSIONS: mNGS has an excellent accuracy in etiological diagnosis and pathogen detection of suspected pneumonia in critically rheumatic patients, which has potential significance for clinical decision-making. Its superiority to different types of pathogens depends on the comprehensiveness of CMTs.