Identification of respiratory microbiota markers in ventilator-associated pneumonia

Abstract

Purpose: To compare bacteria recovered by standard cultures and metataxonomics, particularly with regard to ventilator-associated pneumonia (VAP) pathogens, and to determine if the presence of particular bacteria or microbiota in tracheal and oropharyngeal secretions during the course of intubation was associated with the development of VAP.

Methods: In this case-control study, oropharyngeal secretions and endotracheal aspirate were collected daily in mechanically ventilated patients. Culture and metataxonomics (16S rRNA gene-based taxonomic profiling of bacterial communities) were performed on serial upper respiratory samples from patients with late-onset definite VAP and their respective controls.

Results: Metataxonomic analyses showed that a low relative abundance of Bacilli at the time of intubation in the oropharyngeal secretions was strongly associated with the subsequent development of VAP. On the day of VAP, the quantity of human and bacterial DNA in both tracheal and oropharyngeal secretions was significantly higher in patients with VAP than in matched controls with similar ventilation times. Molecular techniques identified the pathogen(s) of VAP found by culture, but also many more bacteria, classically difficult to culture, such as Mycoplasma spp. and anaerobes.

Conclusions: Molecular analyses of respiratory specimens identified markers associated with the development of VAP, as well as important differences in the taxa abundance between VAP and controls. Further prospective trials are needed to test the predictive value of these markers, as well as the relevance of uncultured bacteria in the pathogenesis of VAP.

Keywords: Etiology; Metataxonomics; Molecular; Pathogenesis; Pneumonia; Prevention; VAP.

Fig. 1

Study flow chart. ICU intensive care unit, MV mechanical ventilation, VAP ventilator-associated pneumonia

Fig. 2

Microbiota composition and diversity change during intubation. The analysis was based on the dataset normalized to 1000 reads per sample. Twenty groups compared were defined by three factors: development of VAP (VAP, control), sample type (OPS, ETA) and sampling point (D0, D3, DVAP-3, DVAP and DVAP + 3). a Mean relative abundances for seven major bacterial phyla (top panel) and Shannon diversity index (bottom panel). b Principal coordinates analysis of the Bray–Curtis similarity matrix based on the square-root transformed relative abundance of bacterial genera. Bacterial communities defined by the same combination of the three factors were grouped to the centroid. Dotted lines connect neighboring time point centroids. Vectors of Pearson correlation (with length > 0.4) between genera and PCo axes are shown above the plot. In control patients, baseline (D0) ETA and OPS microbial communities were significantly different (PERMANOVA p < 0.05) from those of DVAP-3, DVAP and DVAP + 3. In patients who developed VAP, OPS microbiota was significantly different from those of DVAP and DVAP + 3. c–d Changes per patient between D0 and DVAP bacterial communities along PCo1 and PCo2 axes. Changes are represented by dashed lines, in red for patients who developed VAP and in blue for control patients. The change along PCo2 was significantly lower for OPS of patients who developed VAP (P = 0.02, Wilcoxon rank-sum test). Day 0 samples of all patients were arbitrarily placed at the origin of both axes. OPS oropharyngeal secretions, ETA endotracheal aspirate, VAP ventilator-associated pneumonia, D0 intubation day, D3 3 days after intubation, DVAP-3 3 days before VAP, DVAP day of VAP diagnosis, DVAP + 3 3 days after VAP

Fig. 3

Microbial communities from different airway sites of the same patient are closely related. The three sample types (OPS, ETA and BAL) were compared whenever available at exactly the same day for each patient. The group average hierarchical clustering of samples (represented at the top) was based on the Bray–Curtis similarity of square-root-transformed relative abundance of bacterial families. Samples from the same patients are represented on the dendrogram by the patient number and color. Two sets of samples from patient 239 were available for the analysis; they were collected at days 3 and 7 of intubation (indicated in parenthesis). Sample types are indicated by gray symbols below patient numbers. The heat plot (bottom panel) shows the square-root-transformed proportions of bacterial families according to the intensity gradient on the right. OPS oropharyngeal secretions, BAL bronchoalveolar lavage fluid, ETA endotracheal aspirate, VAP ventilator-associated pneumonia

Fig. 4

VAP is associated with taxonomic composition of the microbiota and DNA load in respiratory samples. a Dot plots of the relative abundance of 16S rRNA reads assigned to the class Bacilli in OPS samples at D0. b ROC curve for associations between VAP and the relative abundance of class Bacilli in OPS at D0. c Dot plots depicting human DNA load in extracts of ETA at DVAP. d ROC curve for associations between VAP and bacterial or human DNA load at DVAP in ETA or OPS. The area under the curve, specificity, sensitivity and optimal cut-off are shown for each model in the same color as the ROC curve. ROC receiver operating characteristic, OPS oropharyngeal secretions, ETA endotracheal aspirate, VAP ventilator-associated pneumonia, D0 intubation day, DVAP day of VAP diagnosis

Fig. 5

CAP performed on OPS microbiota at D0 allows a reasonable distinction of patients who develop or not VAP. Results of the CAP for OPS and ETA samples from the first three sampling points (D0, D3, DVAP-3) are presented. The analyses were based on the relative abundance of genera in the dataset normalized to 1000 reads per sample. Bars represent the percentage of correct allocations of samples to VAP and control groups obtained by cross-validation. n number of samples included in the analysis, m number of PCo axes used for the discriminant analysis, ETA endotracheal aspirate, OPS oropharyngeal secretions, VAP ventilator-associated pneumonia, CAP Canonical Analysis of Principal Coordinates, D0 intubation day, D3 3 days after intubation, DVAP-3 3 days before VAP