Pulmonary Microbiome of Patients Receiving Mechanical Ventilation: Changes Over Time

Abstract

Background

Interest in the pulmonary microbiome is growing, particularly in patients undergoing mechanical ventilation.

Objectives

To explore the pulmonary microbiome over time in patients undergoing prolonged mechanical ventilation and to evaluate the effect of an oral suctioning intervention on the microbiome.

Methods

This descriptive subanalysis from a clinical trial involved a random sample of 16 participants (7 intervention, 9 control) who received mechanical ventilation for at least 5 days. Five paired oral and tracheal specimens were evaluated for each participant over time. Bacterial DNA from the paired specimens was evaluated using 16S rRNA gene sequencing. Bacterial taxonomy composition, α-diversity (Shannon index), and β-diversity (Morisita-Horn index) were calculated and compared within and between participants.

Results

Participants were predominantly male (69%) and White (63%), with a mean age of 58 years, and underwent mechanical ventilation for a mean of 9.36 days. Abundant bacterial taxa included Prevotella, Staphylococcus, Streptococcus, Stenotrophomonas, and Veillonella. Mean tracheal α-diversity decreased over time for the total group (P = .002) and the control group (P = .02). β-Diversity was lower (P = .04) in the control group (1.905) than in the intervention group (2.607).

Conclusions

Prolonged mechanical ventilation was associated with changes in the pulmonary microbiome, with the control group having less diversity. The oral suctioning intervention may have reduced oral-tracheal bacterial transmission.

Knowledge regarding roles of microbiomes in human disease is evolving.¹ Microbiomes are collections of microbes at a location.² Microbiomes play active roles in physiology, metabolism, and immune processes.³ Related research has focused on understanding microbiome structure and function,^2,4 its contribution and response to illness and injury,^3,5–7 and diagnostic and therapeutic uses.^8,9 Microbiome perturbations (dysbiosis) are associated with acute and chronic illnesses.^10–12

Microbiome Characterization

Microbiome composition can be determined by high-throughput DNA sequencing of the 16S rRNA marker gene.¹³ Microbial DNA is isolated from a specimen, and the bacterial 16S rRNA gene is amplified using polymerase chain reaction (PCR). Sequence data are generated from the PCR amplicons and used to infer taxonomic composition. Similar sequences are “binned” into operational taxonomic units (OTUs). Each OTU representative sequence is matched to a database of known 16S rRNA sequences.^9,13

Microbiome community structure is described in terms of α- and β-diversity.¹³ α-Diversity is a summary statistic reflecting the number and distribution of taxa within a specific sample (eg, Shannon index). Higher α-diversity is consistent with healthier microbiomes. β-Diversity is a similarity score reflecting overlap between samples (eg, Morisita-Horn index). It reflects the shared taxa across specimens compared with the total number of specimen taxa.

Pulmonary Microbiome Determinants

The pulmonary microbiome is determined by microbial immigration, microbial elimination, and cell reproduction conditions.² Critical illness and associated treatments influence microbiome determinants,^2,3,10 and dysbiosis is common among patients receiving mechanical ventilation.^6–9

Microbial immigration is movement of microbes into the respiratory tract and is influenced by microaspiration and atmospheric air intake,^14,15 which are enhanced by body positioning, level of consciousness, and gastroesophageal reflux.⁵ Microbial elimination is movement of microbes out of the respiratory tract facilitated by mucociliatory escalator activity and coughing,² which are impaired by endotracheal tubes and sedation.^5,16 Cell reproduction conditions are environmental elements promoting or inhibiting cell reproduction, including pH, temperature, perfusion, and immune cell activity, all of which are altered by critical illness and associated interventions.¹⁵

The purpose of this pilot study was to explore the pulmonary microbiome in patients experiencing prolonged mechanical ventilation and to evaluate the effect of an oral suctioning intervention on the microbiome.

Methods

Design

This prospective, descriptive subanalysis was part of a randomized clinical trial (NO-ASPIRATE) designed to evaluate the effectiveness of an oral suctioning protocol in reducing aspiration and ventilator-associated conditions among patients undergoing mechanical ventilation.^17,18 The parent study was registered in the ClinicalTrials.gov database (ID: NCT02284178).

Setting and Sample

A random sample of 16 participants (7 intervention and 9 control) with a minimum mechanical ventilation duration of 5 days was selected from the parent study population. Five paired tracheal and oral specimens collected during the course of mechanical ventilation were selected per participant for analysis: baseline (specimen pair 1), last (specimen pair 5), and 3 intermediate specimens (specimen pairs 2–4).

Tracheal Specimen Sampling Procedures

Trained research assistants obtained paired oral and tracheal specimens every 12 hours for the duration of mechanical ventilation.¹⁷ Standard laboratory protocols were followed to maintain specimen integrity.

Outcome Measures

Bacterial taxonomy composition, α-diversity, and β-diversity were calculated and compared within and between participants.

Tracheal and Oral Microbiome Analysis

Tracheal and oral supernatant and pellets were subjected to genomic DNA extraction (QIAGEN). The DNA was quantified, and a quality check was performed by PCR. The purified genomic DNA was used to amplify the V3-V4 region of the bacterial 16S rRNA gene, covering both conserved and variable regions. The Illumina MiSeq sequencing platform was used to generate read sequence data from the amplicons. After demultiplexing, sequence data were processed using VSEARCH.¹⁹ Processing included merging of forward and reverse reads and removal of low-quality and chimeric sequences. Resulting sequences (ie, good reads) were clustered at 97% sequence identity to generate OTUs. Taxonomic assignments for OTU representative sequences were computed using the RDP (Ribosomal Database Project) classifier implemented in the mothur program.^20,21 Samples with reads below 1000 were discarded.

Statistical Analyses

Participants’ characteristics were summarized using descriptive statistics. The OTU and bacterial taxonomy calculations were performed using R (vegan package).²² Bacterial taxonomy identification, Shannon index, and Morisita-Horn index were determined.²³ Shannon index values across the 5 samples were compared using Friedman 2-way analysis of variance with a Bonferroni correction. Pairwise sample distances based on taxonomic profiles were estimated using the Morisita-Horn index and compared using an independent t test with Welch correction (IBM SPSS Statistics, version 26).

Results

Clinical Data and Patient Groups

Most participants were male (n = 11, 69%) and White (n = 10, 63%), with a mean age of 58 years, a mean Acute Physiology and Chronic Health Evaluation II score of 24.50, and a mean of 9.36 days receiving mechanical ventilation. Demographic characteristics did not differ significantly between groups.

Microbiome Analysis

A total of 10 663 380 good reads were generated and analyzed from the 160 oral and tracheal samples (median, 24 352; mean, 66 646; minimum, 1942; maximum, 2 654 519). High variability was noted in taxonomic compositions of oral and tracheal samples and across groups. The bacterial taxa (genus level) with the highest relative abundances were Prevotella, Staphylococcus, Streptococcus, Stenotrophomonas, and Veillonella.

Changes in α-Diversity Over Time

The Shannon index values were as follows: median, 3.06; mean, 2.99; minimum, 0.02; maximum, 5.86 (see Table).

Table.

α-Diversity changes in oral and tracheal specimens over time

α-Diversity	Total group (N = 16)	Control group (n = 9)	Intervention group (n = 7)
Mean oral ranks
Shannon 1	3.31	3.56	3.00
Shannon 2	3.88	3.67	4.14
Shannon 3	2.38	2.22	2.57
Shannon 4	2.88	2.89	2.86
Shannon 5	2.56	2.67	2.43
Pa	.05	.26	.27
Mean tracheal ranks
Shannon 1	4.06	4.00	4.14
Shannon 2	3.56	3.56	3.57
Shannon 3	2.69	3.22	2.00
Shannon 4	2.69	2.44	3.00
Shannon 5	2.00	1.78	2.29
Pa	.002	.02	.07

^aFriedman 2-way analysis of variance by rank.

In the total group, mean tracheal Shannon diversity values decreased significantly (P = .002) over time (see Figure). A post hoc significant difference was noted between Shannon 1 and Shannon 5 (4.06 vs 2.00, P = .002). Mean oral Shannon diversity values decreased over time, though not significantly (P = .05).

Figure.

Tracheal Shannon diversity comparison over time for the total group (N = 16). Friedman 2-way analysis of variance by rank yielded a P of .002.

In the intervention group, mean oral and tracheal Shannon diversity values decreased over time, though not significantly (P = .27 and P = .07, respectively).

In the control group, mean tracheal Shannon diversity values significantly decreased over time (P = .02). Post hoc analysis showed a significant difference between Shannon 1 and Shannon 5 (4.00 vs 1.78, P = .03). Mean oral Shannon diversity values decreased over time, though not significantly (P = .26).

Transmission of Microbiota From Oral Secretions Into Tracheal Airways

To identify the source of the tracheal microbiome and examine the possibility of oral-tracheal microbiota transmission, β-diversity was separately computed for the control (n = 9) and intervention (n = 7) samples. Mean distance comparison showed a significant difference between the intervention and control groups (2.607 vs 1.905, respectively; 1-tailed P = .04).

Discussion

Changes over time in the pulmonary microbiome of patients receiving mechanical ventilation have been relatively unexplored. Our findings of significant reductions in diversity over time were consistent with those of other published studies.^16,24 A unique feature of our study was the duration of mechanical ventilation (mean, 9.36 days).

Initial tracheal specimens in both groups showed heterogeneity. Over the course of mechanical ventilation, diversity decreased, with Prevotella, Staphylococcus, and Pseudomonas becoming dominant. Eventually, tracheal specimens were dominated by 1 or 2 specific taxa, which is consistent with the findings of other studies.²⁴ In this small pilot study, tracheal α-diversity decreased significantly in the control group but not in the intervention group. The parent trial’s suctioning intervention may have contributed to maintaining a more diverse pulmonary microbiome.²⁵

Evaluation of the oral microbiome initially showed heterogeneity. Diversity in both the intervention and control groups decreased over time, but not significantly. Common organisms included Prevotella, Stenotrophomonas, and Staphylococcus.

Seven participants (44%) experienced ventilator-associated conditions. One patient (control group) experienced an infection-related ventilator-associated condition, and cultures revealed Streptococcus pneumoniae and Staphylococcus aureus.²⁶

Antibiotic use is common in patients receiving mechanical ventilation and may contribute to changes in oral and tracheal microbiomes.⁵ In our sample, 2 patients (12%) received no antibiotics, and 4 patients (25%) received antibiotics every day. The proportion of antibiotic-free days was similar in the 2 groups (P = .92).

Conclusion

In this study sample, pulmonary bacterial diversity decreased significantly during the course of mechanical ventilation. Tracheal diversity decreased significantly in participants in the control group but not in the intervention group. These preliminary findings suggest that the oropharyngeal suctioning intervention was effective in reducing the oral-tracheal transmission of bacteria. Further exploration of the pulmonary microbiome during mechanical ventilation is needed.