# Co-occurrence network analysis reveals novel associations between the neonatal airway microbiome and bronchopulmonary dysplasia risk: an observational, population-based study

## TL;DR

The study finds that the neonatal airway microbiome's network structure is linked to the severity of bronchopulmonary dysplasia in preterm infants.

## Contribution

The study introduces microbial network stability as a novel risk factor for severe bronchopulmonary dysplasia.

## Key findings

- Network complexity in the airway microbiome decreases with increasing BPD severity.
- Keystone taxa like Acinetobacter and Fusobacterium are associated with different BPD grades.
- Higher microbial network density at birth is linked to a reduced risk of BPD.

## Abstract

To evaluate the association between respiratory tract microorganisms at birth and the subsequent development and severity of bronchopulmonary dysplasia (BPD) in preterm infants. This prospective cohort study enrolled 98 preterm infants (gestational age < 32 weeks, birth weight < 2,000 g). Tracheal aspirate samples were collected through endotracheal intubation within 2 h after birth. Using 16S rRNA sequencing, we characterized the airway microbiome and performed co-occurrence network analysis with compositionally robust methods. Among 98 preterm infants analyzed, the incidence of BPD was 68.4%, comprising 31 grade I, 20 grade II, and 16 grade III cases. Airway microbiota in infants with BPD exhibited distinct severity-stage patterns: Escherichia-Shigella and Streptococcus were significantly enriched in grade I, while Chryseobacterium increased markedly in grade III, accompanied by a significant reduction in Streptococcus. Microbial co-occurrence network analysis yielded three key insights. (i) Network complexity declined sharply with BPD severity, being sparsest in grade III. (ii) Distinct keystone taxa were identified across different groups: Acinetobacter and Fusobacterium in the non-BPD group; Brevundimonas and Fusobacterium in grade I; Fusobacterium and Acinetobacter at grade II and grade III. (iii) In a multivariable model adjusted for key clinical confounders, a higher microbial network density at birth was independently associated with a substantially reduced risk (OR = 0.12, P < 0.05). The ecological architecture of the neonatal airway microbiome at birth, defined by network complexity and keystone taxa, is associated with BPD severity. This highlights microbial network stability as a novel factor and ecological interactions as a target for future research.

Bronchopulmonary dysplasia (BPD) remains the most common chronic lung disease in preterm infants. While its pathogenesis is incompletely understood, the role of the early respiratory microbe is increasingly recognized. Previous studies have largely focused on individual pathogenic taxa, overlooking the complex ecological interactions within microbial communities. Our analysis reveals that the architecture of microbial co-occurrence networks in the neonatal airway varies significantly with BPD severity. Notably, network complexity decreased markedly as BPD severity increased. We identified specific keystone taxa uniquely associated with disease outcomes, suggesting that microbial ecosystem stability rather than individual species may be a critical factor in BPD pathogenesis. These findings shift the focus from single microbes to the stability of the microbial ecosystem as a novel risk factor for severe BPD, offering new avenues for risk stratification and early intervention.