Metagenomic next-generation sequencing of bronchoalveolar lavage fluid from children with severe pneumonia in pediatric intensive care unit

Lower respiratory tract infections (LRTIs), particularly severe pneumonia, pose a significant threat to children's health globally, causing substantial morbidity and mortality. Accurate and timely identification of the causative pathogens is crucial for effective treatment. Traditional diagnostic methods, such as microbial cultures, often suffer from limitations in sensitivity and speed, potentially delaying appropriate interventions. The rise of antibiotic resistance further complicates the situation, underscoring the need for rapid and sensitive diagnostic tools. This study focuses on the application of metagenomic next-generation sequencing (mNGS) as a novel approach for identifying pathogens in children with severe pneumonia admitted to a PICU. mNGS offers a comprehensive, culture-independent method to detect a wide range of pathogens, including bacteria, viruses, and fungi, directly from clinical samples. Bronchoalveolar lavage fluid (BALF) is a particularly suitable sample for detecting respiratory pathogens, offering a higher sensitivity than other methods like sputum or blood cultures. Previous studies have demonstrated the efficacy of mNGS in identifying pathogens in adults with severe pneumonia, but its application in children remains relatively unexplored. Therefore, this study aims to characterize the microbiome of BALF in children with severe pneumonia using mNGS, identify potential pathogens associated with the disease, and explore correlations between microbial composition and clinical indicators. The results of this study can potentially improve the accuracy and speed of diagnosis, leading to more timely and effective treatment strategies for children with severe pneumonia in the PICU, ultimately improving patient outcomes and reducing mortality.

Several studies have demonstrated the potential of mNGS in diagnosing respiratory infections in adults. Leo et al. (2017) established a protocol for extracting DNA from BALF samples for mNGS. Fang et al. (2018) successfully used mNGS to diagnose adenovirus and herpes simplex virus encephalitis. Qi et al. (2019) showed mNGS's ability to detect pathogens in culture-negative ventilator-associated pneumonia cases. Li et al. (2020a) compared mNGS sensitivity and specificity to culture and PCR methods, showing improved sensitivity. Liu et al. (2019) and Kalantar et al. (2019) compared mNGS results from BALF and other samples, highlighting the method's strengths and limitations depending on the sample type. However, studies focusing on children with severe pneumonia are limited. Takeuchi et al. (2019) conducted mNGS on BALF from children with respiratory failure, identifying pathogens missed by conventional methods. This study builds upon these findings by examining a larger cohort of children with severe pneumonia in a PICU setting, analyzing a broader spectrum of microbial communities, and correlating the mNGS results with clinical parameters.

This retrospective study enrolled 126 children diagnosed with severe pneumonia admitted to the PICU of Children's Hospital of Fudan University between February 2018 and February 2020. Severe pneumonia was defined using WHO criteria, including invasive mechanical ventilation, fluid-refractory shock, need for non-invasive ventilation, and severe hypoxemia. Exclusion criteria included age <28 days or >18 years, non-infectious causes of respiratory distress, contraindications to bronchoscopy, and severe cardiopulmonary dysfunction. BALF samples were collected using an age-appropriate pediatric flexible fiberoptic bronchoscope following established guidelines. Approximately 1 mL/kg of warm saline (max 20 mL) was instilled, and at least 40% was recovered by suction. Serum samples were collected for inflammatory marker analysis (lipopolysaccharide (LPS), procalcitonin (PCT), C-reactive protein (CRP), interleukin 6 (IL-6)) and lymphocyte subpopulation analysis (B cells, CD3+ T cells, CD4+ T cells, CD8+ T cells, NK cells) using flow cytometry. Conventional microbiological cultures were performed. mNGS was performed on 126 BALF samples, with DNA sequencing for most samples (101/126) and RNA sequencing for a subset (25/126). Data analysis included normalization of sequencing reads, calculation of microbial diversity indices (Chao1, Shannon), and correlation/regression analysis between microbial diversity and clinical parameters using R software. Pathogenic microorganisms were determined based on relative abundance and z-score distribution, referring to a previous study by Yan et al. (2021).

mNGS identified 88 potentially pathogenic bacterial species, 13 viral species, 39 fungal species, and 27 protozoan species in the BALF samples. The most abundant bacterial species included various Streptococcus and Staphylococcus species and Acinetobacter species. Detected viruses included four human herpesviruses (Human alphaherpesvirus 1, Human cytomegalovirus, Human betaherpesvirus 6B, Epstein-Barr virus), Human mastadenovirus B, and Human polyomavirus 4. Fungal species included Aspergillus species and Pneumocystis jirovecii. Protozoa included Plasmodium and Trypanosoma species. The BALF bacterial diversity index (Chao1 and Shannon) showed a positive correlation with serum inflammatory markers (LPS, CRP, PCT, IL-6) and lymphocyte subtypes. The abundance of specific bacteria (e.g., Pseudomonas aeruginosa) correlated with specific inflammatory markers and inversely with WBC and CD4+ T cell counts. Viral abundance in BALF was positively correlated with maximum body temperature and negatively correlated with CD3+ lymphocyte count. The abundance of Human betaherpesvirus 6B and Human gammaherpesvirus 4 showed a negative correlation with total WBC count. Human cytomegalovirus abundance was significantly higher in immunodeficient patients. Higher fungal and protozoan richness (Chao1) and diversity (Shannon) in BALF were significantly associated with mortality. The abundance of certain bacteria (Escherichia coli, Klebsiella pneumoniae, Corynebacterium segmentosum), fungi (Pneumocystis jirovecii), and viruses (Human mastadenovirus B, Torque teno virus 29) correlated positively with the oxygenation index (OI), a measure of pneumonia severity. BALF fungal and parasite diversity was significantly higher in patients with sepsis. The total number of bacterial species in BALF was significantly higher in immunodeficient patients. Acinetobacter baumannii and Streptococcus mitis were significantly more abundant in hospital-acquired pneumonia (HAP) than community-acquired pneumonia (CAP), while Schaalia odontolytica was less abundant in HAP.

This study demonstrates the utility of mNGS in identifying a diverse range of potential pathogens in children with severe pneumonia. The correlations observed between microbial diversity and inflammatory markers and immune cell populations support the role of these microbes in the pathogenesis of severe pneumonia. The findings highlight the potential for co-infections with viruses and fungi, particularly in immunodeficient children. The association of increased fungal and protozoan diversity with mortality underscores the importance of considering these organisms in managing severe cases. The limitations of relying solely on culture-based methods are highlighted by the detection of pathogens not identified by conventional methods. While the study did not directly guide treatment decisions based on mNGS results due to the turnaround time at the time, the current ability to obtain results within 24 hours allows for real-time clinical decision-making.

This study demonstrates the effectiveness of mNGS in identifying diverse pathogens in children with severe pneumonia. The correlations identified between the microbiome and clinical parameters provide valuable insights into disease pathogenesis. Future research should focus on validating the mNGS findings using other methods, investigating the role of specific pathogens in disease severity, and determining the clinical impact of incorporating mNGS into routine diagnostic practices. Larger multicenter studies are also warranted to confirm these findings and establish mNGS as a valuable tool for improving clinical outcomes in pediatric severe pneumonia.

The study's retrospective nature and the limited BALF volume available for inflammatory marker analysis are limitations. The findings rely on identifying potential pathogens based on sequence data, requiring further validation through culture and antibody testing. The correlation between mNGS findings and clinical indicators are observational and do not establish causal relationships. The study population was limited to a single PICU, potentially affecting the generalizability of the results.