557 - Ecological continuity of bacterial communities in the respiratory and gastrointestinal tracts in ventilator dependent infants with bronchopulmonary dysplasia (BPD)

Poster Number: 557
Publication Number: 557.336

Erik A. Jensen, Children's Hospital of Philadelphia, haddonfield, NJ, United States; Ceylan Tanes, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Krithika Lingappan, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Joseph Piccione, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Stamatia Alexiou, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Julie L. Fierro, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Clyde Wright, University of Colorado School of Medicine, Denver, CO, United States; G. Scott Worthen, Childrens Hospital of Philadelphia, Philadelphia, PA, United States

Background: Scant data characterize the respiratory and gastrointestinal (GI) microbiota in infants with BPD.

Objective: To use molecular methods to define the respiratory and GI bacterial microbiota in ventilator dependent preterm infants with BPD.

Design/Methods: Subjects underwent bronchoalveolar lavage (BAL), tracheal aspirate (TA), nasal swab, oral swab, gastric fluid aspiration, and stool collection. Bacterial 16S rRNA gene V1-V2 region was sequenced using Illumina MiSeq. Sequence data were processed using QIIME2. Community-level differences were assessed by weighted UniFrac distances. BAL neutrophil count and subject respiratory severity score (RSS) defined lung inflammation and BPD acuity.

Results: Ten infants were studied (Table). Library microbial DNA concertation was highest in oral and stool samples (Fig 1A). Bronchoscope and TA pre-wash samples differed significantly (by DNA library count and weighted UniFrac distances) from BAL and TA samples, indicating scant environmental contamination (Fig 1B). In 9 of 10 infants, Proteobacteria were the most abundant organisms detected by 16S sequencing of BAL fluid (Neisseria n=6, Pseudomonas n=2, Enterobacteriaceae n=1)(Fig 2A). BAL alpha diversity was not associated with BAL neutrophil count or RSS, albeit with a possible decrease in Shannon index with greater neutrophils (p=.09). Within subject comparisons showed similar weighted UniFrac distances between BAL and TA fluid (p=.14; n=10) and BAL and gastric fluid (p=.12; n=3)(Fig 2B), but significant differences between BAL and all other sampling loci (p < .05). The most abundant organism was consistent in paired BAL and TA samples in 90% of infants (Fig 2A). Similarity between BAL and TA fluid increased (UniFrac distances decreased) with greater BAL neutrophil count (p < .05) but not RSS (p=.32) (Fig 2C). Bacteria were cultured from BAL fluid in 7 of 10 infants. The same cultured organisms were detected at the genus level by 16S sequencing of BAL fluid in 6 of 7 infants. However, the genera identified by culture were the most abundant by 16S in only 2 of 7 infants and genera relative abundance varied widely (1%-99.6%).Conclusion(s): In ventilator dependent infants with BPD, the microbiota of the respiratory and GI tracts differed significantly between most of the sampled anatomical loci. However, our data support TA collection as an alternative to BAL for microbiome analyses in BPD but suggest BAL fluid culture and 16S sequencing may yield discordant results.
Table. Subject characteristics and bronchoalveolar lavage (BAL) results
Figure 1.(A) Mean (standard deviation shown in error bars) library DNA concentration detected in respiratory and gastrointestinal samples, bronchoscope and tracheal aspirate catheter pre-washes, and other control samples. (B) Principal coordinate analysis (PCoA) plot showing pairwise weighted UniFrac distances between BAL fluid, tracheal aspirate fluid, and bronchoscope pre-wash, and tracheal aspirate catheter pre-wash. Plot shows distinct relationships between patient respiratory vs. equipment pre-wash samples and higher DNA library concentrations in respiratory vs. pre-wash samples.