Optimisation and clinical validation of a metagenomic third-generation sequencing approach for aetiological diagnosis in bronchoalveolar lavage fluid of patients with pneumonia

This is the authors' abstract. We don't add key points for this paper.

BACKGROUND: Metagenomic Third Generation Sequencing (mTGS), based on nanopore technology, has emerged as a promising tool for the rapid diagnosis of pneumonia pathogens. However, this technology currently lacks standardised technical protocols, quality control measures, and comprehensive performance evaluations for the simultaneous detection of bacteria, fungi, and viruses in clinical settings. METHODS: We optimised the mTGS workflow by refining key parameters (cell wall lysis, fragment size selection, host DNA depletion, and sequencing depth) using reference samples and bronchoalveolar lavage fluid (BALF) from eight patients with pneumonia. These optimisations formed the basis for a standardised mTGS protocol. To assess the clinical diagnostic value of the optimised mTGS, a multicentre prospective cohort study involving 313 pneumonia-suspected patients was conducted. Each BALF sample was tested using conventional microbiological testing (CMTs), metagenomic next-generation sequencing (mNGS), pre-optimised mTGS, and optimised mTGS. FINDINGS: The optimised mTGS protocol, based on the refined cell wall lysis, fragment size selection, no host DNA depletion, and 800 MB sequencing depth, achieved a tenfold increase in sensitivity compared with pre-optimised mTGS for detecting the species of Bacillus subtilis, Mycobacterium tuberculosis, Mycobacterium avium, Cryptococcus neoformans, and Human papillomavirus in reference samples. In the prospective cohort, 274 patients with a confirmed diagnosis of pneumonia were identified, yielding 376 distinct microbes. The mTGS identified more microbes than CMTs (314 vs. 115), with a 45.30% increase in sensitivity (84.70% vs. 39.40%, P < 0.01, Chi-square test/Fisher's exact test). Compared with pre-optimised mTGS, the sensitivity of optimised mTGS increased by 32.51% (84.70% vs. 52.19%, P < 0.01, Chi-square test/Fisher's exact test). mTGS showed comparable performance to mNGS (84.70% vs. 79.90%, P = 0.14,Chi-square test/Fisher's exact test), both significantly outperforming CMTs. mNGS was more sensitive for detecting Non-tuberculous mycobacteria, Pneumocystis jirovecii, and Aspergillus spp., while mTGS demonstrated higher sensitivity for M. tuberculosis, Chlamydia psittaci, and Streptococcus pneumoniae. The overall diagnostic agreement between mTGS and clinical diagnosis was 81.80%. INTERPRETATION: We optimised and validated a standardised mTGS protocol that significantly improved the ability to detect pathogens in the BALF of patients with pneumonia. Optimised mTGS demonstrated comparable performance to mNGS, making it a promising tool for the aetiological diagnosis of pneumonia.