Discussion
This was the first study to identify bacterial compositional differences
between patients with asthma and healthy controls using EBC-derived EVs.
Patients with asthma exhibited significantly different diversity and
richness indicators compared to the healthy controls. Five bacterial
genera, including Sphingomonas , Akkermansia ,Methylophaga, Acidocella , and Marinobacter were
significantly abundant in the EBC EVs of patients with asthma, whereasAcinetobacter , Staphylococcus , Bifidobacterium ,Blautia , and Collinsella were more abundant in the
controls. No significant difference was observed between the EBC
microbiota in eosinophilic and non-eosinophlic asthma. We suggested a
diagnostic tool for asthma using AI modeling based on the distinctness
of the bacterial community. Firmicutes and Proteobacteria were common
important features at the phylum level in the GBM and ANN asthma models.
This metagenomic analysis of EBC-derevied EVs enhances the understanding
of the bacterial communities residing in the airway. EVs are composed of
a lipid bilayer that contains transmembrane proteins which affect host
cells and other bacteria, even those located far from the colonized area17,18. Moreover, they participate in intercellular
communication and have potential applications as biomarkers and
therapeutic agents for several allergic diseases19,20. Airway studies focusing on the microbiota and
EVs in nasal lavage and lung tissue samples reported that EVs have
distinct characteristics associated with chronic rhinitis and chronic
obstructive pulmonary diseases (COPD) 18,21.
Although the advances in metagenomics have elucidated the lung
microbiota, sampling the lower respiratory tract is a challenge.
Bronchoscopy enables investigation of the lower airway with greater
precision, but is invasive. The use of bronchial brushes and
bronchoalveolar lavage (BAL) fluid is disputed due to potential
contamination from the oropharynx 22. The collection
of induced sputum is noninvasive; however, it requires preliminary
inhalation of an aerosol and is often impractical. The EBC collection
technique is a safe that can be repeated easily, and the sample contains
nonvolatile water-soluble compounds from the airway mucosa, which are
likely to reflect the composition of the airway lining fluid23,24. Therefore, we used EBC-derived EVs to identify
the bacterial composition of patients with asthma for the first time.
The relationship between asthma and lung bacterial community diversity
is controversial. Previous studies in patients with asthma have
documented increased bacterial burden and diversity using induced sputum
and bronchial brushing compared with healthy controls2,25. Similarly, other studies reported an increase in
the bacterial diversity in BAL samples obtained from 23 patients with
eosinophil-low asthma 26. This study also showed
greater bacterial diversity in patients with asthma based on Chao1,
Shannon, and Simpson indices compared with controls using EVs derived
from EBC. Conversely, other studies failed to replicate these findings
and occasionally presented opposite results. A study reported that there
were no significant differences in alpha diversity between patients with
severe and non-severe asthmatics, and healthy controls27,28. This discrepancy may be attributed to the
differences in the populations from which samples were obtained and in
the sampling collection methods and sites.
In this study, EBC EV samples obtained from patients with asthma had a
significantly higher proportion of Sphingomonas ,Akkermansia and Methylophaga at the genus level compared
to the controls. Sphingomonas species have been associated with
increased airway hyperresponsiveness to methacholine and eosinophil-high
asthma 2,26. Firmicutes and Proteobacteria were phyla
that were demonstrated to be increased in the patients with asthma in
the GBM ad ANN asthma models. A previous study reported that relevant
differences were found in Firmicutes and Proteobacteria between induced
sputum samples obtained from asthmatics and healthy
controls25, which is consistent with our results.
Other studies using bronchial samples reported that Proteobacteria were
present in higher proportions in patients with asthma compared to the
controls 2,3. Elevated levels of the Firmicutes phylum
have been found in the lung tissue-derived EVs of patients with COPD,
which may be explained by differences in pH, oxygenation levels, and
temperature of the lung environment related to esophageal reflux18. Patients with chronic lung diseases such as asthma
and COPD share similar bacteria causing dysbiosis, such as Protebacteria
and Firmicutes29. Although we can speculate on the
causal relationship between the microbiota and airway inflammation, we
cannot confirm its existence yet. Further studies are needed to
investigate the specific causal relationship between the microbiota and
airway inflammation mediated via the molecular pathways. Additionally,
based on the result of the microbiome-based differences between patients
with asthma and controls, we proposed a diagnostic AI model of asthma
with good performance. This might be a biomarker for risk assessment and
treatment of asthma through further studies.
As the scope of microbiome-focused research widens, numerous studies
have reported distinct respiratory microbiota according to the various
asthma phenotypes and endotypes, such as obesity, severity, airway
hyperresponsiveness, asthma control, acute exacerbation, and response to
steroids 30-34. In particular, the eosinophilic
inflammatory phenotype and underlying T2-high endotype corresponded to
lung microbial communities with comparatively lower bacterial loads and
substantial diversity 35. Unfortunately, we did not
demonstrate the distinction in the airway microbiota according to blood
eosinophil counts in this study. Our results may be attributed to the
relatively small number of EBC samples and variations in the
participants, which may be potentially attributable to populations,
environment, and antibiotic use history.
Recent immunomodulatory treatments for asthma mainly focused on the
mechanisms of type 2 inflammation, such as biologics, i.e., antibodies
against cytokine mediators (e.g., IL-5, IL-4, and
IL-13)31. However, these treatments are not effective
for patients with a T2-low endotype, which may limit the indications of
biologics in them. The strong association between airway dysbiosis and
pathogenesis of neutrophilic asthma has also been reported36,37; thus, the manipulation of the airway and gut
microbiomes may be a novel therapeutic strategy for asthma, especially
in patients with T2-low endotypes. Further investigation into the impact
of microbiota therapies may be a useful tool for controlling asthma
using a personalized medicine approaches.
This study has several limitations. Firstly, the sample size of both
groups was relatively small. Larger studies are needed to confirm our
finding. Second, baseline characteristics such as dietary patterns and
use of antibiotics were not evaluated in controls and asthmatics, which
might have affected the bacterial community as a confounding bias.
Nevertheless, to the best our knowledge, this was the first study to
analyze the microbiota using EBC-derived EVs in patients with asthma and
compare it with healthy controls.
In conclusion, we have demonstrated differences between the lung
microbiome of patients with asthma and controls by analyzing EVs derived
from EBC. We also proposed an asthma diagnostic model using AI. Our
findings suggest that asthma may be associated with dysbioisis of the
bacterial community, and the microbiome obtained using EBC EVs may be a
potential indicator for the diagnosis of asthma. Further studies are
needed to verify the distinct patterns of the microbiome according to
asthma phenotypes and endotypes, considering the heterogeneity of
asthma.