Diagnostic performance of the metagenomic next-generation sequencing in lung biopsy tissues in patients suspected of having a local pulmonary infection.

PURPOSE

This study aims to evaluate the diagnostic application and performance of the metagenomic next-generation sequencing (mNGS) in patients suspected of local pulmonary infection by comparing it to the traditional pathogen detection methods in lung tissue specimens obtained by a computerized tomography-guided biopsy (CT-guided biopsy).

METHODS

We retrospectively reviewed patients, admitted to the First Affiliated Hospital of Wenzhou Medical University, China from May 2018 to December 2020, who were suspected of local pulmonary infection. All cases received a CT-guided lung biopsy, tissue samples were sent both for conventional examinations (CE) and mNGS tests. The sensitivity and specificity of the two diagnostic approaches were compared.

RESULTS

106 patients enrolled, 76 patients were diagnosed with a pulmonary infection. Among 49 patients with identified pathogens, CE confirmed pathogenic infections in 32 cases. Mycobacterium spp. and fungi accounted for 37.5% (12/32) and 28.1% (9/32), respectively, with bacteria 34.4% (11/32). The mNGS examination detected extra pathogenic microorganisms in 22 patients that were consistent with the patients' clinical and radiographic pictures. The sensitivity of mNGS was 53.9% vs. 42.1% for the CE, while the specificity was 56.7% versus 96.7%. For detection rate, mNGS was significantly superior to CE in bacterial (96.3% vs. 40.7%, p < 0.05), and mixed infections (100% vs. 50%, p < 0.05), but inferior to CE in fungal (60% vs. 90%, p > 0.05) and Mycobacterium spp. infections (66.7% vs. 100%, p > 0.05) with no significant difference. Among 31 cases diagnosed with lung abscess, the diagnostic performance of the detection rate was 67.7% (21/31) in favour of mNGS compared to 29.0% (9/31) for CE (p < 0.05). Most polymicrobial infections were induced by anaerobic species that coexisted with Streptococcus constellatus. And Klebsiella pneumoniae was the most common isolated monomicrobial infection.

CONCLUSIONS

The most commonly detected causative pathogens for local pulmonary infections were bacteria, Mycobacterium spp. and fungi. Compared with the CE, the advantages of mNGS in the pathogens detection lie in the discovery of bacterial and mixed infections, as well as in the detection of lung abscess. Conversely, mNGS is not good enough to be recommendable for the detection of Mycobacterium spp. and fungi.