1. Introduction
Infectious diseases remain most common causes of death among all patient populations worldwide [1]. Pulmonary infections lead to more deaths each year than any other infectious disease categories [2]. Delayed identification of etiologic pathogens in pulmonary infections is the key cause of treatment failure and death. Current microbiological tests such as culture-based methods often fail to identify the etiologic pathogens in most cases of pulmonary infection, due largely to the limitations in terms of sensitivity [3, 4], speed and spectrums of available assay targets [5]. In these situations, failing to identify the etiology microorganisms often promotes the initiation of empiric antibiotic therapy, which may lack activity against an underlying microorganism, leading to treatment failure, disease progression, and consequent adverse outcomes [5, 6].
Swift identification of causative microorganisms and tailoring of the antimicrobial regimen is highly desirable, which would improve the prognosis of pulmonary infections due to optimization of antimicrobial treatment. Recent rapid advances in genomic sequencing techniques and bioinformatics has made it possible for metagenomic next-generation sequencing (mNGS) to be used in clinical diagnostics. [7 ~ 9]. mNGS is an unbiased approach that can theoretically detect all microorganisms in a clinical sample [10], which offers potential alternatives for detecting etiologic microorganisms and distinguish them from background commensal microorganisms, which might pave the way for personalized medicine [11]. Several reports have attempted to detect pathogens using mNGS in infectious diseases, such as central nervous system infections [12 ~ 14], digestive infections [15, 16], bloodstream infections [7, 18] and pulmonary infections [19 ~ 23]. Published information on the use of mNGS in diagnosing pulmonary infections is sparse, and the types of samples tested by mNGS were mainly limited to percutaneous lung biopsy samples [19] and bronchoalveolar lavage fluid (BALF) samples [20 ~ 23].
Accurate sampling of peripheral pulmonary lesions (PPLs), which were defined as lesions surrounded by normal lung parenchyma presents challenges and were unlikely to be visualized by bronchoscopy, is the key step. CT-guided percutaneous needle biopsy has a high diagnostic yield for PPIs, but is more invasive and with a relatively high incidence of complications, such as bleeding or pneumothorax [24, 25]. Bronchoscopy remains an appropriate initial investigation for PPLs, due to the lower complication rate [26]. Most previous studies have focused on the use of bronchoscopy in diagnosing malignancy. There is little published information on the use of bronchoscopy in diagnosing infectious PPLs.
The present study consists of two parts: on the one hand, we retrospectively evaluate the performance of mNGS for samples from infectious PPLs obtained using ultrathin bronchoscopy in conjunction with virtual bronchoscopic navigation (VBN) and rapid on-site cytological evaluation (ROSE), and compare diagnostic accuracy of mNGS with conventional cultures; on the other hand, we evaluate the comparative diagnostic performance and microbial composition from different sampling including transbronchial lung biopsy (TBLB), bronchial needle brushing (BB) and BALF with mNGS.