Lung microbiota of patients with ARDS due to coronavirus disease 2019 receiving ECMO

Abstract

Diversity of the microbiota, which is essential for lower airway homeostasis, is greatly altered in acute respiratory distress syndrome (ARDS). Extracorporeal membrane oxygenation (ECMO) is the ultimate protective treatment for the lungs of patients with severe ARDS, but little is known about its effect on the lung microbiota of these patients. To evaluate the effect of ECMO on the lung microbiota of ARDS patients, we performed 16S rRNA and fungal ITS1 profiling and shotgun sequencing on bronchoalveolar lavage fluid (BALF) collected from ARDS patients due to COVID-19. BALF was collected from 13 patients, five of whom underwent ECMO. In all patients, Pseudomonas was the most abundant of the bacteria. The patients with ECMO had more Pseudomonas and more Klebsiella than those without ECMO. The most abundant fungi were unspecified fungi in the patients with ECMO and Emmia lacerata in the patients without ECMO. Alpha diversity of bacteria and fungi did not differ significantly between the two groups. Human betaherpesvirus 5 and human alphaherpesvirus 1 were predominant in all patients, with human betaherpesvirus 5 decreasing over time in the ECMO patients. The patients with ARDS due to COVID-19 receiving ECMO had a different lung microbiota than those not receiving ECMO.

Introduction

The lungs of healthy adults contain a unique microbial community that is involved in the maintenance of respiratory physiology and immune homeostasis by forming a network of ecological interaction between the microbiota and the host^1–3. Microorganisms abundant in the lower respiratory tract include the Firmicutes (Streptococcus and Veillonella) and Bacteroidetes (Prevotella) among bacteria; Aspergillus, Candida, Cladosporium, Malassezia, and Saccharomyces among fungi; and Anelloviridae and Redondoviridae among the viruses^4–7. Dysbiosis, defined as deviation from the normal microbial composition, is involved in the development and progression of respiratory disease^8,9. In patients with acute respiratory distress syndrome (ARDS), lung diversity decreases over time and dysbiosis progresses due to both the progression of the disease itself and the effects of positive pressure ventilation.

Extracorporeal membrane oxygenation (ECMO) is a treatment for patients with severe ARDS in which gas-exchanged oxygenated blood is delivered by an extracorporeal artificial lung to replace inadequate tissue oxygen supply for days to weeks and to prevent the development of ventilator-induced lung injury due to exposure to high concentrations of oxygen and hyperventilation associated with ventilation. Thus, the lungs are temporarily rested to prevent irreversible damage to the injured lung while it is being treated and restored¹⁰. To our knowledge, there are no reports to date on the lung microbiota of ARDS patients treated with ECMO.

The purpose of this study was to determine the effect of ECMO on the lung microbiota of patients with ARDS due to COVID-19.

Materials and methods

Study design and participants

This single-center, prospective, observational clinical study was conducted in patients with ARDS due to COVID-19 who were admitted directly from the emergency department to the Division of Trauma and Surgical Critical Care, Osaka General Medical Center between April 2021 and March 2022. The diagnosis of COVID-19 was confirmed by polymerase chain reaction testing of nasal swabs on admission. The diagnosis of ARDS followed the Berlin definition¹¹. All patients with ARDS were treated in accordance with established clinical guidelines, including lung-protective ventilation strategies such as low tidal volume ventilation and appropriate PEEP settings¹². In addition, patients were managed with an internal suction system as part of routine ventilator care. Clinical and biological parameters such as patient demographic characteristics, duration of mechanical ventilation and hospitalization, and comorbidities were collected from the electronic medical record. Severity scores were recorded using the Acute Physiology and Chronic Health Evaluation (APACHE) II score (range 0–71) and Sequential Organ Failure Assessment (SOFA) score (range 0–24), with ARDS severity rated as mild, moderate, or severe^13,14. This study was approved by the institutional review board of Osaka General Medical Center (approval number: 2021-002). Written informed consent was obtained from all patients. The study was conducted in accordance with the Declaration of Helsinki.

Sample collection

Bronchoalveolar lavage fluid (BALF) was collected from all patients using a bronchial fiberscope. In the patients without ECMO, BALF was collected within 24 h after the diagnosis of ARDS. In the patients with ECMO, BALF was collected within 24 h after ECMO initiation and as needed after the initial BALF collection. The bronchoalveolar lavage procedure was performed under aseptic conditions using a disposable AMBU® ASCOPE™ 4 (Ambu A/S, Ballerup, Denmark). Bronchoalveolar lavage was performed according to standardized procedures, specifically by injecting 3 × 20 mL of sterile saline solution into the bronchi. After each injection, the largest volume of fluid in the bronchioles (nearly 10 mL total) was collected in 50 mL sterile plastic tubes and centrifuged to separate the supernatant from the sediment. The tubes were stored at − 80 °C until use.

Amplicon library construction and sequencing

Bacterial and fungal DNA were extracted from the precipitate fraction of BALF using a PI-1200 nucleic acid extraction system (Kurabo, Osaka, Japan). For bacterial metagenome analysis, the V1-V2 variable region of the 16S rRNA gene was sequenced in 251-bp paired-end mode on an Illumina MiSeq platform (Illumina, San Diego, CA, USA). For fungal metagenome analysis, the fungal ITS1 region was sequenced in 301-bp paired-end mode on the Illumina MiSeq. The resulting paired-end sequences were merged, filtered, and denoised using DADA2 software (https://benjjneb.github.io/dada2/). Taxonomic assignments were made using the QIIME2 feature-classifier plug-in in the Greengenes database (release 13_8) for bacteria and the ntF-ITS1 database for fungi. The QIIME2 pipeline, version 2020.2, was used as the bioinformatics environment for processing all relevant raw sequence data.

Metagenomic shotgun sequencing

Viral RNA was extracted from the precipitate fraction using a QIAamp MinElute Virus Spin Kit (Qiagen, Hilden, Germany). The extracted RNA was then used to synthesize double-stranded DNA using a ProtoScript II First Strand cDNA Synthesis Kit (New England Biolabs, Ipswich, MA, USA), NEBNext Ultra II Non-Directional RNA Second Strand Synthesis Module (New England Biolabs), and Random Primer 6 (random hexanucleotides; New England Biolabs). Next, viral metagenome shotgun libraries were prepared for each sample using a Twist Library Preparation Enzymatic Fragmentation Kit (Twist Bioscience, South San Francisco, CA, USA) and the Twist Comprehensive Viral Research Panel (Twist Bioscience). All libraries were converted to libraries for DNBSEQ using a MGIEasy Universal Library Conversion Kit (App-A). Sequencing was performed using a DNBSEQ-G400RS High-throughput Sequencing Kit (MGI Tech, Tokyo, Japan) in 100-bp paired-end mode. Each read was subjected to Kraken2 analysis against the PlusPFP database, which includes archaeal, bacterial, viral, plasmid, human, UniVec_core, protozoan, fungal, and plant sequences.

Statical analysis

Continuous variables are shown as the median and interquartile range (IQR), and categorical variables are shown as frequencies and percentages. The Wilcoxon rank-sum test was used to test continuous variables, and Fisher’s exact test was used to test the nominal variables. A p-value of < 0.05 was considered to indicate statistical significance. All analyses were performed using JMP Pro17 (SAS Institute Inc., Cary, NC, USA) and Prism 9 (GraphPad Software, Boston, MA, USA).

Results

Patient characteristics

In total, 13 patients were included in the study: 5 patients with ECMO and 8 patients without ECMO. Patient characteristics are shown in Table 1. The median age (IQR) of the patients with ECMO was significantly younger than that of those without ECMO (44 [36–48] vs. 64 [53–74] years, p = 0.007). The median APACHE II score of the patients with ECMO was significantly higher than that of those without ECMO (20 [17–22] vs. 15 [12–18], p = 0.018). The median SOFA score was also significantly higher than that of those without ECMO (10 [9–13] vs. 8 [4–9], p = 0.022). None of the patients received non-invasive respiratory support, such as high flow nasal cannula oxygen or non-invasive positive pressure ventilation, before undergoing invasive mechanical ventilation. As adjunctive therapy for ARDS, 80% of the patients without ECMO were treated in the prone position, whereas all of the patients with ECMO were treated in the lateral position. The median length of ECMO was 11 (10–22) days. There was no significant difference in the duration of mechanical ventilation between the patients with ECMO and those without ECMO. The median length of stay in the intensive care unit (ICU) was significantly longer in the patients with ECMO (13 [11–29] vs. 9 [6–12] days, p = 0.045). There was no significant difference in mortality between the two groups. The mechanical ventilation parameters used within 24 h of ARDS diagnosis for both groups are shown in Supplementary Table S1 online. The patients with ECMO had extremely low ventilator volume settings.

Table 1.

Characteristics of the population.

	ECMO (−)	ECMO (+)	Total	P value
	n = 8	n = 5	N = 13
	Sample n = 8	Sample n = 13	Sample N = 21
Demographics
Age (years), median (IQR)	64 (53–74)	44 (36–48)	51 (44–70)	0.007
Sex, male (%)	6 (75)	5 (100)	11 (84.6)	0.487
BMI, median (IQR)	28.6 (24.8–30.5)	33.7 (20.3–36.9)	28.7 (24.6–32.2)	0.558
Current smoker	0 (0)	2 (40)	2 (15.4)	0.128
Former smoker	3 (37.5)	1 (20)	4 (30.8)	0.506
Comorbidities, n (%)
Hypertension	2 (25)	1 (20)	3 (23.1)	0.835
Diabetes	1 (12.5)	1 (20)	2 (15.4)	0.715
Immunocompromise	1 (12.5)	0 (0)	1 (7.7)	0.411
Cardiovascular compromise	0 (0)	0 (0)	0 (0)	NA
Chronic obstructive pulmonary disease	0 (0)	0 (0)	0 (0)	NA
Renal insufficiency	0 (0)	0 (0)	0 (0)	NA
COVID-19 vaccination status, vaccinated, n (%)	4 (50)	0 (0)	4 (30.8)	0.105
Days after onset, days, median (IQR)	7 (4–8)	12 (7–17)	7 (6–12)	0.055
Severity of disease on admission
APACHE II score, median (IQR)	15 (12–18)	20 (17–22)	17 (14–19)	0.018
SOFA score, median (IQR)	8 (4–9)	10 (9–13)	8 (6–10)	0.022
Severity of ARDS, n (%)				0.231
Severe	5 (62.5)	5 (100)	10 (76.9)
Moderate	3 (37.5)	0 (0)	3 (23.1)
Mild	0 (0)	0 (0)	0 (0)
Treatment of disease, n (%)
Antibiotics	2 (25)	2 (40)	4 (30.8)	0.571
Tazobactam and piperacillin	0 (0)	2 (40)	2 (15.4)	0.487
Ampicillin and sulbactam	1 (12.5)	0 (0)	1 (7.7)	0.385
Trimethoprim and sulfamethoxazole	1 (12.5)	0 (0)	1 (7.7)	0.385
Antivirals	3 (30)	3 (60)	6 (46.2)	0.592
Lopinavir	1 (12.5)	0 (0)	1 (7.7)	0.411
Favipiravir	1 (12.5)	0 (0)	1 (7.7)	0.411
Remdesivir Immunomodulators	2 (25)	3 (60)	5 (38.5)	0.293
Glucocorticoid	8 (100)	5 (100)	13 (100)	NA
Tocilizumab	0 (0)	0 (0)	0 (0)	NA
Baricitinib	0 (0)	0 (0)	0 (0)	NA
Adjunctive therapies for ARDS, n (%)
Neuromuscular blockade	8 (100)	5 (100)	13 (100)	NA
Prone position	8 (80)	0 (0)	8 (61.5)	< 0.001
Lateral position	0 (0)	5 (100)	5 (38.5)	< 0.001
Disease course
Length of ECMO, days, median (IQR)	0 (0)	11 (10–22)	11 (10–22)
Length of mechanical ventilation, days, median (IQR)	12 (6–20)	12 (11–29)	12 (8–21)	0.27
Length of stay in ICU, days, median (IQR)	9 (6–12)	13 (11–29)	10 (7–16)	0.045
Length of stay in hospital, days, median (IQR)	23 (11–27)	30 (22–39)	25 (16–33)	0.107
ICU mortality	0 (0)	1 (20)	1 (7.7)	0.385
Hospital mortality	3 (37.5)	1 (20)	4 (30.8)	0.506
Number of days from admission date of sample collection
Within 24 h	8	5
2–7 days	0	3
7–14 days	0	2
> 14 days	0	3

ARDS, acute respiratory distress syndrome; IQR, interquartile range; BMI, body mass index; APACHE, Acute Physiology and Chronic Health Evaluation; SOFA, Sequential Organ Failure Assessment; ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit; NA, not available.

Lung bacterial microbiota

In all samples, Proteobacteria and Firmicutes were predominant in the composition of flora at the phylum level (Fig. 1A). Figure 1B shows the top bacterial phyla with relative abundance greater than 1% in the ECMO and non-ECMO groups. The top three of these bacterial phyla were the same in both groups, but their frequencies differed: the relative frequencies of Proteobacteria, Firmicutes, and Actinobacteria in the non-ECMO group were 51.3%, 29.3%, and 6.3%, whereas in the ECMO group they were 76.6%, 8.0%, and 6.5%. The composition of the bacterial flora at the genus level in all samples is shown in Fig. 1C. The top bacterial genera averaging over 1% relative abundance in the ECMO and non-ECMO groups are shown in Fig. 1D. Pseudomonas was the most predominant in both groups. The top three Gram-negative rods (Pseudomonas, Curvibacter, and Sphingomonas) tended to be more common in the ECMO group, whereas Streptococcus tended to be higher in the non-ECMO group (Fig. 1E). The relative changes in abundance of the top six bacterial genera in patients with ECMO are shown in Fig. 1F. Pseudomonas, Curvibacter, and Sphingomonas showed similar trends within the same patient. In two patients, Klebsiella tended to increase through day 21. There was little similarity in beta diversity between the patients with and without ECMO and in terms of the number of days on mechanical ventilation (Fig. 2). The patients with ECMO did not differ in alpha diversity from those without ECMO (see Supplementary Fig. S1 online).

Fig. 1.

Lung bacterial microbiota in the patients with and without ECMO. (A) Composition of the lung microbiota at the phylum level for the samples collected within 24 h of ARDS diagnosis. The stacked bars indicate the mean relative abundance at the portal level of all samples. (B) Composition of the top lung microbiota at the phylum level for the samples collected within 24 h of ARDS diagnosis. Bar graph shows the top bacterial phyla in the groups with and without ECMO by averaging the bacterial phyla with a relative abundance of 1% or greater. (C) Composition of the lung microbiota at the genus level for the samples collected within 24 h of ARDS diagnosis. The legend indicates the top 20 genera. (D) Composition of the top lung microbiota at the top genus level for the samples collected within 24 h of ARDS diagnosis. Bar graph shows the top bacterial genera in the groups with and without ECMO by averaging the bacterial genera with a relative abundance of 1% or greater. (E) Box-and-whisker diagrams show the relative abundance of the top four genera in samples collected within 24 h of ARDS diagnosis in the patients with and without ECMO. (F) The relative changes in abundance of the top six bacterial genera in the patients with ECMO. ECMO, extracorporeal membrane oxygenation.

Fig. 2.

PCoA2D plots of β diversity analysis of lung bacterial flora in the patients with and without ECMO. (A) PCoA2D plots of β diversity analysis for BALF collected within 24 h of ARDS diagnosis from patients with and without ECMO. Dissimilarity between samples was measured by unweighted UniFrac distance. (B) PCoA2D plots of β diversity analysis for BALF by days of mechanical ventilation in all patients. Dissimilarity between samples was measured by unweighted UniFrac distance. < 1 day indicates within the first day of mechanical ventilation. 2 days–7 days indicates from the second day to the seventh day of mechanical ventilation. > 7 days indicates from day 7 onward of mechanical ventilation. (C) PCoA2D plots of β diversity analysis for BALF collected within 24 h of ARDS diagnosis from patients with and without ECMO. Dissimilarity between samples was measured by weighted UniFrac distance. (D) PCoA2D plots of β diversity analysis for BALF by days of mechanical ventilation in all patients. Dissimilarity between samples was measured by weighted UniFrac distance. < 1 day indicates within the first day of mechanical ventilation. 2 days-7 days indicates from the second day to the seventh day of mechanical ventilation. > 7 days indicates from day 7 onward of mechanical ventilation. BALF, bronchoalveolar lavage fluid; ECMO, extracorporeal membrane oxygenation; PCoA2D, principle coordinate analysis 2-dimensional; ANOSIM, analysis of similarities.

Lung mycobiota

The composition of the fungal flora at the species level in all samples is shown in Fig. 3A. Each sample was dominated by only a few fungal species. The top fungal species averaging over 1% relative abundance are shown in Fig. 3B. In the ECMO group, unclassified fungi were the most predominant, followed in abundance by Malassezia restricta. In the non-ECMO group, Emmia lacerata was the most predominant, followed in abundance by M. restricta. The patients with ECMO tended to have less diverse lung mycobiota than those without ECMO in terms of the Shannon and Simpson diversity indexes (see Supplementary Fig. S2 online). There were no differences in beta diversity based on ECMO status or number of days on mechanical ventilation (Fig. 3C).

Fig. 3.

Lung mycobiota at the species level in the patients with and without ECMO. (A) Composition of lung mycobiota at the species level for the samples collected within 24 h of ARDS diagnosis. The legend indicates the 27 species. (B) Composition of the top lung mycobiota at the species level for the samples collected within 24 h of ARDS diagnosis. Bar graph shows top fungal species in the groups with and without ECMO by averaging the fungal species with a relative abundance of 1% or greater. (C) PCoA2D plots of β diversity analysis of lung fungal flora. PCoA2D plots of β diversity analysis for BALF collected within 24 h of ARDS diagnosis in patients with and without ECMO. Dissimilarity between samples was measured by unweighted UniFrac distance. (D) PCoA2D plots of β diversity analysis of lung fungal flora. PCoA2D plots of β diversity analysis for BALF by days of mechanical ventilation in all patients. Dissimilarity between samples was measured by unweighted UniFrac distance. < 1 day indicates within the first day of mechanical ventilation. 2 days-7 days indicates from the second day to the seventh day of mechanical ventilation. > 7 days indicates from day 7 onward of mechanical ventilation. (E) PCoA2D plots of β diversity analysis of lung fungal flora. PCoA2D plots of β diversity analysis for BALF collected within 24 h of ARDS diagnosis in patients with and without ECMO. Dissimilarity between samples was measured by weighted UniFrac distance. (F) PCoA2D plots of β diversity analysis of lung fungal flora. PCoA2D plots of β diversity analysis for BALF by days of mechanical ventilation in all patients. Dissimilarity between samples was measured by weighted UniFrac distance. < 1 day indicates within the first day of mechanical ventilation. 2 days-7 days indicates from the second day to the seventh day of mechanical ventilation. > 7 days indicates from day 7 onward of mechanical ventilation. BALF, bronchoalveolar lavage fluid; ECMO, extracorporeal membrane oxygenation; PCoA2D, principle coordinate analysis 2-dimensional; ANOSIM, analysis of similarities.

Lung virome

The composition of viral flora in all samples is shown in Fig. 4A. A bar graph of the viruses averaged by relative abundance greater than 1% in each patient group with and without ECMO is shown in Fig. 4B. All patients had COVID-19, but human betaherpesvirus 5 and human alphaherpesvirus 1, not severe acute respiratory syndrome coronavirus, were predominant. ECMO patients tended to have a lower relative abundance of human betaherpesvirus 5 and a greater relative abundance of severe acute respiratory syndrome coronavirus. The changes in relative abundance of the top five viruses in the patients with ECMO are shown in Fig. 4C. The relative abundance of human betaherpesvirus 5 tended to decrease over the 21 days, but it increased in one patient after this time.

Fig. 4.

Lung virome at the species level in the patients with and without ECMO. (A) Composition of the lung virome at the species level for samples collected within 24 h of ARDS diagnosis. The legend indicates the top 20 species. (B) Composition of the top lung virome at the species level for samples collected within 24 h of ARDS diagnosis. Bar graph showing the top viruses by averaging the species with a relative abundance of 1% or greater in the patients with or without ECMO. (C) The relative changes in abundance of the top five viruses in patients with ECMO. ECMO, extracorporeal membrane oxygenation.

Discussion

We profiled the bacterial, fungal, and viral flora of the lower respiratory tract of patients with ARDS due to COVID-19. The microbiota of the lower respiratory tract of patients with ECMO was rich in Pseudomonas among the bacteria, unclassified fungi among the fungi, and human betaherpesvirus 5 (Cytomegalovirus: CMV) among the viruses.

The Proteobacteria phylum is predominant in the lower respiratory tract of patients with ARDS due to COVID-19, and an increase in the abundance of the Pseudomonas genus and Enterobacter has been reported as a characteristic of the lower respiratory tract of patients with pneumonia due to COVID-19, which is consistent with the results of the present study^15,16. Notably, patients with ECMO tended to have higher relative abundances of Pseudomonas and Klebsiella than those without ECMO, whereas patients without ECMO tended to have higher relative abundances of Streptococcus and Staphylococcus aureus. Viral infections damage tissues of the respiratory tract pathway, leading to dysbiosis and the promotion of bacterial colony formation¹⁷. In COVID-19, Pseudomonas aeruginosa has been reported to promote colony formation, and enrichment of P. aeruginosa is associated with poor prognosis^18,19. Enrichment of Pseudomonas may have been greater in the ECMO patients with severe ARDS, who have more severe lung injury. The oral microbiota of the elderly is rich in staphylococci and streptococci, and an increased relative abundance of Enterobacter in the gut has been reported in patients with severe ARDS^20–22. Aspiration in patients with ARDS also affects the microbiota of the lower respiratory tract^23,24. Thus, there may be an increase in streptococci in patients without ECMO and, because of the severity of the disease, a greater confirmed presence of Enterobacter predominantly in the ECMO patients. Despite the enrichment and severity of highly pathogenic microorganisms in the ECMO patients, clinical outcomes and diversity were not inferior to those of the non-ECMO patients. ECMO replaces oxygenation and carbon dioxide removal that the lungs would normally do and allows for protective ventilation that significantly reduces plateau and driving pressures²⁵. This results in a significant reduction in the concentrations of plasma sRAGE, interleukin-6, and monocyte chemotaxis protein-1, thus limiting pulmonary biotrauma caused by mechanical ventilation²⁶. ECMO has also been reported to promote recovery of alveolar epithelial function in rat experiments²⁷. The fact that changes in lung microbiota did not lead to a decrease in diversity in the ECMO patients may have contributed to the protective effect of ECMO on the lungs.

The fungal flora of ARDS patients has been reported to be enriched with Candida albicans, and C. albicans is a risk factor for death^28,29. This was similarly observed in ARDS due to COVID-19, in which an increase in unidentified ascomycetes in ARDS patients who were not contaminated with Candida spp. was reported³⁰. This is consistent with the results of the present study, which showed patients contaminated with C. albicans from an early stage and an increase in unidentified fungi in the uncontaminated cases. Notably, M. restricta was abundant in the present study. Malassezia restricta is endemic to the skin and intestinal tract³¹, and in the intestinal fungal flora, M. restricta and C. albicans are prominent³². Similar to the bacterial flora, M. restricta may have been enriched through the gut-lung axis in ARDS patients.

The respiratory virus flora are thought to play an important role in the pathogenesis of respiratory disease by interacting with the immune system^7,33. Tobacco mosaic virus is reported to be abundant in COVID-19 and both Anelloviridae and Redondoviridae are abundant in severe cases, and the presence of these viruses is positively correlated with intubation during hospitalization^34,35. In the present study, there was marked enrichment of CMV and Human alphaherpesvirus 1 (Herpes simplex virus 1: HSV-1) but not of Anelloviridae and Redondoviridae. The patients with ECMO tended to have less Human betaherpesvirus 5 than those without ECMO. Human betaherpesvirus 5 and Human alphaherpesvirus are the most commonly identified viruses from patients on mechanical ventilation, and viral reactivation among ICU patients, especially among the herpes group, significantly changes the virome³⁶. It was reported that CMV pulmonary infections, not HSV-1, were associated with increased length of mechanical ventilation and increased ICU length of stay and mortality in ventilated patients³⁷. Given the decreasing trend in the relative abundance of CMV with the increasing duration of mechanical ventilation in ECMO patients, the lung rest provided by ECMO may contribute to the suppression of pathogenic viral enrichment. The subjects of the previously reported studies were early in the initiation of mechanical ventilation, and there are no reports on the progression of single-virus enrichment with increased duration of mechanical ventilation.

Limitations

This study has several limitations. First, several confounding factors were not considered. It has been reported that enrichment of highly pathogenic bacteria is associated with disease progression and severity³⁸. The ECMO patients were significantly more severely ill than the non-ECMO patients, which may have resulted in enrichment of highly pathogenic bacteria due to the severity of the disease. In addition, treatment prior to sample collection may have affected the lung microbiota. Antibiotics have been reported to alter the lung microbiota and reduce diversity³⁹. It is possible that antibiotic therapy might have influenced the results of this study. We found negative factors for the lung microbiota such as severity of illness and enrichment of highly pathogenic bacteria in the ECMO patients, but the lack of reduced diversity and worsening outcomes may have been due to the lung-protective effect of ECMO. The present results are those of a preliminary study based on measurements of a very small number of cases at a single center. Validation in larger numbers of cases and in other cohorts would be needed to accurately assess the impact of ECMO on the lung microbiota. Second, although we detected a large number of microorganisms using metagenomic sequencing, contamination in the airways and of bronchoscopes should always be considered. We used disposable bronchoscopes to minimize contamination. Third, we did not evaluate whether the microorganisms were truly pathogenic or only present in the airways. Further studies using animal models should provide important insights into the pathogenic role of microbiome alterations in patients with ECMO.

Conclusion

The ARDS patients treated with ECMO had a different lung microbiota than those without ECMO. It is speculated that critical illness, respiratory management, and a variety of other factors contribute to the lung microbiota of ARDS patients. Further studies are needed to determine how the unique respiratory management of ECMO affects the lung microbiota.

Abbreviations

APACHE II

Acute physiologic assessment and chronic health evaluation II

ARDS

Acute respiratory distress syndrome

BALF

Bronchoalveolar lavage fluid

CMV

Cytomegalovirus

COVID-19

Coronavirus disease 2019

ECMO

Extracorporeal membrane oxygenation

HSV-1

Herpes simplex virus 1

ICU

Intensive care unit

IQR

Interquartile range

SOFA

Sequential organ failure assessment

Author contributions

Y.M. designed the study, analyzed the data, and wrote the manuscript; D.M. performed sequencing and helped analyze the data; K.S., H.O., S.F., and J.O. critically revised the manuscript for intellectual content. All authors read and approved the final manuscript.

Funding

This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science [Grant Number 22K09132].

Data availability

The datasets used and analyzed in the study can be found in online repositories (accession number PRJDB17654). This data can be found here: https://ddbj.nig.ac.jp/resource/bioproject/PRJDB17654.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

This study was approved by the institutional review board of Osaka General Medical Center (approval number: 2021-002). Written informed consent was obtained from all patients. The study was conducted in accordance with the Declaration of Helsinki.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-08664-1.