Application of Metagenomic Next-Generation Sequencing in the Diagnosis of Pulmonary Infectious Pathogens From Bronchoalveolar Lavage Samples

Background: Metagenomic next-generation sequencing (mNGS) is a powerful method for pathogen detection. In this study, we assessed the value of mNGS for bronchoalveolar lavage (BAL) samples in the diagnosis of pulmonary infections. Methods: From February 2018 to April 2019, BAL samples were collected from 235 patients with suspected pulmonary infections. mNGS and microbial culture were performed to evaluate the effectiveness of mNGS in pulmonary infection diagnosis. Results: We employed mNGS to evaluate the alpha diversity, results suggesting that patients with confirmed pathogens had a lower microbial diversity index compared to that of patients with uncertain pathogens. For the patients admitted to the respiratory intensive care unit (RICU) or on a ventilator, they experienced a lower diversity index than that of the patients in the general ward or not on a ventilator. In addition, mNGS of BAL had a diagnostic sensitivity of 88.89% and a specificity of 14.86% in pulmonary infection, with 21.16% positive predictive value (PPV) and 83.87% negative predictive value (NPV). When rare pathogens were excluded, the sensitivity of mNGS decreased to 73.33%, and the specificity increased to 41.71%. For patients in the simple pulmonary infection group and the immunocompromised group, the main infection types were bacterial infection (58.33%) and mixed-infection (43.18%). Furthermore, mNGS had an advantage over culture in describing polymicrobial ecosystem, demonstrating the microbial distribution and the dominant strains of the respiratory tract in patients with different underlying diseases. Conclusions: The study indicated that mNGS of BAL samples could provide more accurate diagnostic information in pulmonary infections and demonstrate the changes of respiratory microbiome in different underlying diseases. This method might play an important role in the clinical use of antimicrobial agents in the future.