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

Abstract

Background: Severe pneumonia due to lower respiratory tract infections (LRTIs) is a significant cause of morbidity and mortality in children. Noninfectious respiratory syndromes resembling LRTIs can complicate the diagnosis and may also make targeted therapy difficult because of the difficulty of identifying LRTI pathogens. In the present study, a highly sensitive metagenomic next-generation sequencing (mNGS) approach was used to characterize the microbiome of bronchoalveolar lavage fluid (BALF) in children with severe lower pneumonia and identify pathogenic microorganisms that may cause severe pneumonia. The purpose of this study was to use mNGS to explore the potential microbiomes of children with severe pneumonia in a PICU.

Methods: We enrolled patients meeting diagnostic criteria for severe pneumonia admitted at PICU of the Children's Hospital of Fudan University, China, from February 2018 to February 2020. In total, 126 BALF samples were collected, and mNGS was performed at the DNA and/or RNA level. The pathogenic microorganisms in BALF were identified and correlated with serological inflammatory indicators, lymphocyte subtypes, and clinical symptoms.

Results: mNGS of BALF identified potentially pathogenic bacteria in children with severe pneumonia in the PICU. An increased BALF bacterial diversity index was positively correlated with serum inflammatory indicators and lymphocyte subtypes. Children with severe pneumonia in the PICU had the potential for coinfection with viruses including Epstein-Barr virus, Cytomegalovirus, and Human betaherpesvirus 6B, the abundance of which was positively correlated with immunodeficiency and pneumonia severity, suggesting that the virus may be reactivated in children in the PICU. There was also the potential for coinfection with fungal pathogens including Pneumocystis jirovecii and Aspergillus fumigatus in children with severe pneumonia in the PICU, and an increase in potentially pathogenic eukaryotic diversity in BALF was positively associated with the occurrence of death and sepsis.

Conclusions: mNGS can be used for clinical microbiological testing of BALF samples from children in the PICU. Bacterial combined with viral or fungal infections may be present in the BALF of patients with severe pneumonia in the PICU. Viral or fungal infections are associated with greater disease severity and death.

Keywords: Epstein–Barr virus; Pneumocystis jirovecii; bronchoalveolar lavage fluid; metagenomic next-generation sequencing; severe pneumonia.

Figure 1

Distribution of microbial abundance and z-score detected by mNGS of BALF samples. (A) Bacteria; (B) Viruses; (C) Fungi; (D) Protozoa. An rpm value of <0.01 abundance defaults to a log10 (rpm) of −2 for that microorganism in that sample, and a z-score pf >6 defaults to 6 for all samples. The red dots represent potential pathogens in our selected samples.

Figure 2

Heat map of potentially pathogenic bacteria detected by mNGS of BALF samples. For the top 25 bacteria, each horizontal row represents a sample, and each vertical column represents potentially pathogenic bacteria. The plus and minus signs in the graph represent significant positive and negative correlations, respectively. The clinical phenotype is at the top, and the color block on the right represents the value of the specific clinical phenotype.

Figure 3

Correlations of potentially pathogenic bacteria in BALF with serum inflammatory indicators. The volcano plot displays the correlations of the potentially pathogenic bacteria in BALF with the serum (A) LPS, (B) CRP, (C) PCT, and (D) IL-6 concentrations. The horizontal coordinate represents the Spearman correlation coefficient, and the vertical coordinate represents the negative logarithm of the p-value of the correlation between the clinical phenotype and potential pathogen; i.e., −log10 (p value). Each point in the graph represents a potentially pathogenic microorganism, and those screened for significant differences have been marked in red.

Figure 4

Significant differences in pathogenic microbial composition of bronchoalveolar lavage fluid in the CAP and HAP groups. Significant differences were noted in potentially pathogenic (A) bacteria, (B) fungi, and (C) protozoans between patients with CAP versus HAP. *p < 0.05, **p < 0.01.

Figure 5

Ecological diversity and composition of pathogenic microorganisms in BALF associated with mortality. (A) Mortality associated with Chao1; (B) Mortality associated with Shannon index; (C) Mortality associated with species count; (D) Mortality associated bacteria; (E) Mortality associated with fungi. ab, abandoned; de, deceased; non-de, non deceased. *p < 0.05, **p < 0.01.

Figure 6

Correlation of oxygenation index with pathogenic microorganisms in BALF of patients with severe pneumonia. (A) The OI was positively correlated with the abundance of pathogenic bacteria in BALF; (B) The OI was positively correlated with the abundance of pathogenic fungi in BALF; (C) The OI was positively correlated with the abundance of viruses in BALF; (D) The OI was correlated with the overall abundance of viruses (viral abundance) in BALF.