Lung cancer remains the cardinal cause of cancer-related deaths worldwide (1). Main treatment options for lung cancer include surgery, radiotherapy, chemotherapy, driver gene-targeted therapy, and the newly emerged immunotherapy. Patients with early-stage lung cancer usually have satisfactory prognosis after treatments, but most patients have advanced diseases at their first diagnosis and missed the opportunity for curative treatments. The overall 5-year survival rate of lung cancer is still less than 20% (2).

Microbiota refers to all microorganisms that exist in human body sites such as skin, the digestive tract, urogenital tract, and airway. Approximately 10–100 trillion microorganisms live in the human body, and the number of their cells and genes are far more than that of human hosts themselves (3). Microbiota participate in various physiological processes including digestion, metabolism, immune system maturation, defense of pathogens invasion, and the differentiation and gene expression of mucosal epithelium (4). Dysbiosis in the current study refers broadly to disruption of microbial homeostasis with compositional or functional deviation from microbiota seen in healthy people without lung disease (5). As a part of tumor stroma, dysbiosis of commensal microbiota is associated with development of various malignant tumors and about 13% of cancer cases were driven by microbial pathogen. (6). The luminal microorganisms may affect cancer progression by changing the permeability of the mucosal barrier, activating inflammatory pathways, producing bacterial toxins that impair host genome stability, releasing cancer-promoting metabolites, and modulating local immune microenvironment (7).

As the widest mucosa-coated tissue in the human body, the respiratory tract is continuously exposed to the oropharynx, microorganisms in the air, and other damaging factors. The traditional theory based on bacterial culture techniques deems a healthy respiratory tract as sterile. However, high-throughput sequencing of the airway specimens indicates that the respiratory tract has a low biomass and diverse microbiota (8). Compared with gut microbiota, the airway microbiota has distinct microbial structures and microenvironment (8–10). The disturbance of airway microbiota is involved in the pathogenesis of various respiratory diseases including pneumonia, asthma, cystic fibrosis, chronic obstructive pulmonary diseases, and bronchiectasis (11). Recent studies also indicated that the initiation and progression of lung cancer were related to respiratory microbiota (7, 12, 13). Although dysbiosis of gut microbiota is observed in lung cancer patients, the local bacterial burden of the respiratory tract, rather than the gut, is correlated with the progression of lung tumors (13, 14).

Our group previously demonstrated that, compared with benign pulmonary diseases and healthy controls, dysbiosis of respiratory microbiota was observed in lung cancer patients. Certain bacterial taxa might be used as potential biomarkers for lung cancer (10). Here, we extend our work to lung cancer patients with different staging and pathology to examine the association of airway microbiota with clinical characteristics of lung cancer. We hypothesized the presence of distinct airway microbiota among lung cancer patients with different TNM staging and pathological types. The current study also tried to provide preliminary lines of evidence about the potential roles of respiratory microbiota in the pathobiology of lung cancer.

Commensal microbiota, especially those in the gut, have become potential therapeutic targets for various infectious and inflammatory diseases, as well as tumors (4, 7). Although the biomass in the respiratory tract is relatively low, the airway microbiota has been implicated in the evolution of lung cancer (13, 22, 23). However, studies focusing on relationships between respiratory microbiome and pathological types, and TNM staging, in real-world lung cancer patients are scarce. Moreover, among the respiratory microbiome, the specific taxa that participate in the pathogenesis of lung cancer remain unknown. The current study investigated the composition and diversity of respiratory microbiota in patients with benign pulmonary diseases and lung cancer. To our knowledge, this report provides new insight into the lung cancer tissue microbiota in clinical settings.

At the phylum level, similar to the study conducted by Greathouse et al., increased Proteobacteria and decreased Firmicutes and Fusobacterium were observed in lung cancer tissues compared with non-cancer population control lung tissues (24). In contrast, Actinobacteria were less frequent in the lung cancer group (especially the LUSC group). Considering the recent discovery that certain strains of Actinobacteria might produce anticancer metabolites, it is justified to further explore the association between lung cancer pathogenesis and local Actinobacteria activities in the airway (25). At the genus level, Brevundimonas and Limnobacter were abundant in all participants. Brevundimonas strains are Gram-negative, motile bacteria of the family Caulobacteraceae. In humans, Brevundimonas is an opportunistic pathogen able to cause a range of hospital-acquired infections such as bacteremia, pneumonia/pleuritis, urinary tract infection, and skin and soft tissue infection (26). Whether they are associated with cancer remains to be explored. Limnobacter was reported to be one of the potential biomarkers for hepatocellular carcinoma (27). More research on their biological role in the lung and cancer is still needed. Three genera, namely, Streptococcus, Prevotella, and Aggregatibacter, seemed to be enriched in LUSC cases. These genera also played a role in colon, oral cavity, and throat cancer (28–30). In particular, Streptococcus and Prevotella were also found abundant in lung cancer patients in previous studies, and certain Streptococcus-derived proteins could be carcinogenic by inducing epithelial inflammation (10, 28). Moreover, enrichment of Streptococcus and decreased Staphylococcus was found in the respiratory microbiome of lung cancer patients, while studies have shown that Staphylococcus can destroy DNA and Streptococcus plays a role in preventing DNA damage (31, 32). These contradictory findings may be due to the fact that the true carcinogenic taxa have not been discovered, or the same species may play different roles under different pulmonary local environmental conditions. Jungnickel et al. reported that nontypeable H. influenzae-associated inflammation could promote proliferation of KRAS-mutated adenomatous lesions in a TLR-dependent manner (33). However, Haemophilus was less abundant in the LUAD group in the present study, since nontypeable H. influenzae tend to colonize in the airways of patients with COPD. The low incidence of COPD in our LUAD cohort (8.89%) may explain the difference. In future studies, sequencing techniques with higher resolution (e.g., full-length 16S rRNA or metagenome) and Haemophilus culturomics may help elucidate the paradox. Leucothrix and Sulfurovum were mainly found in patients with lung cancer, and their activities in cancer remain unknown. Sphingomonas and Bifidobacterium were more commonly seen in patients with BPD. Increased distributions of Sphingomonas in non-cancerous tissues were similarly indicated in several studies, and coculture of certain strains of Sphingomonas and the fungus Aspergillus fumigatus KMC-901 could produce cancer invasion inhibitory compounds (34–36). More studies are needed to clarify whether Sphingomonas spp. can be cancer-inhibitory. Preclinical models have shown that respiratory microbiota pattern alteration seemed to elicit inflammatory processes by upregulating several cytokines and dysregulating oxidative damage-related biomarkers (37). Oxidative stress is closely related to carcinogenesis. With respect to microbiota modulation, possible ways to combat oxidative stress include local probiotic taxon transplantation, targeting microbiota-derived pro-inflammatory metabolites and so forth. In the future, prospective interventional studies will help to elucidate the specific bacterial taxa, metabolites, and their roles in lung cancer. LEfSe analysis is commonly used to evaluate the association between microbiome and disease status (19, 38). LefSe analysis also showed that Bifidobacterium, Lactobacillus, and Ruminococcaceae, which produce short-chain fatty acids (SCFAs), are enriched in the BPD group of our study. SCFAs have been involved in inflammation alleviation, immune modulation, and anticancer effects (39, 40). Therefore, a reduction of the bacteria that generate SCFAs in the lung cancer tissue may be linked to cancer risk. Moreover, certain bacterial strains like Bifidobacterium spp. have been widely identified inside tumors and are capable of killing tumor cells in a natural or genetically modified form (41). The microbiome has long been reported as a reliable biomarker for the efficacy of immune checkpoint blockade therapy towards lung cancer, and antibiotic-induced microbiota dysbiosis is associated with poorer outcome following immune checkpoint inhibitor treatment (42). Mechanistically, certain bacterial epitopes might enhance the therapeutic effect of immune checkpoint inhibitors by mimicking tumor neoantigens (43). Therefore, knowing the microbiota differences between tumor and benign diseases may help recognize the potentially beneficial probiotics and epitopes, and develop new direct anti-cancer therapies as well as improve the treatment response to immunotherapy among cancer patients.

Common materials used in studies on respiratory microbiota include sputum, bronchoscopy samples, bronchoalveolar lavage fluid (BALF), and lung tissue biopsy. Due to the low-biomass nature of the lung microbiome, the respiratory samples were vulnerable to contamination. The former three types of samples are all at risk of the potential contaminants from the oral cavity. Furthermore, the specimens collected from bronchoscopy or oral sites might ignore the spatial variation of lung microbiota due to the different distribution of microorganisms in the upper and lower respiratory tract (8). Compared with BALF, lung tissue was considered as the preferred specimen to mitigate those confounding interferences and reliably reflect the local microbial composition (44). Several studies investigated the tissue microbiome of lung cancer patients, but most results came from comparisons between tumor samples and tumor-adjacent samples (22, 45–47). Since the airway space is highly dynamic, the contents in the respiratory tract, such as secretions, particles, and cells (including microbiome), are likely to spread through the lumen with the respiratory movement. As a result, the microbial compositions of the tumor-adjacent tissues may also be affected by the tumor microbiota. In this study, the control group tried to reduce those potential contaminations by using surgically removed tissue from BPD patients.

Consistent with the results of previous studies, significant reduction of alpha-diversity in respiratory microbiota of NSCLC patients was observed (13, 19, 45). Our results also revealed that bacterial beta-diversity was different between NSCLC and BPD, and beta-diversity in LUAD samples was also different from that in LUSC. These diversity differences implied that the local environment and bacterial community were different between different lung cancer pathological subtypes. The composition of the respiratory microbiome is mainly regulated by the balance of bacterial immigration, excretion, and replication (48). Reduced diversity and subsequent enrichment of certain microbes in the respiratory tract may lead to carcinogenesis and cancer evolution. For comparisons of respiratory microbial beta-diversity between NSCLC of different TNM stages, differences between stage I and stage III were significant, suggesting that the local airway environments changed with different cancer development stages. Kim et al. indicated that declined alpha-diversity was found in lung cancer of more advanced stages (III–IV). However, no such difference was found in our study (49). Moreover, similar to results of most microbiota studies, whether these microbiota changes are the cause, mediator, or result of cancer evolution and what the specific cancer-promoting component(s) in respiratory microbiota are remain unknown. Due to the limited sample size, more transregional multi-center studies are necessary to further validate those differences.

Antibiotics are considered as broad-spectrum manipulators of commensal microbiota. In the current comparisons of lung cancer, bacterial diversity among recent usage of antibiotics, except for the beta-diversity of respiratory microbiota in lung adenocarcinoma, showed significant difference, and both the alpha-diversity and taxonomic abundances showed similar results across different pathologies. After excluding lung cancer patients with recent antibiotic exposure, more taxa with significantly different abundances were revealed between lung cancer patients and patients with benign pulmonary diseases. Observational studies indicated that antibiotic use was an independent risk factor for lung cancer occurrence. Antibiotic exposure might promote carcinogenesis via disruption of commensal microbiota, which could contribute to dysregulation of host immune homeostasis (50). Our results supported the idea that short-term antibiotic consumption seemed to be a subsidiary rather than a dominant power in molding the composition of airway microbiota. Antibiotics might affect the community structure of respiratory microbiota to some extent. However, because different types and doses of antibiotics may have varied effects on bacteria, more detailed work is necessary to clarify how antibiotics shape the commensal microbiota and whether these changes have a role in lung cancer evolution. Furthermore, current methods for precisely modulating the gut microbiome mainly include fecal microbiota transplantation, probiotics and prebiotics, and diet control. Similar new techniques, such as respiratory microbiota transplantation and bacteriophages, are necessary to elucidate the role of certain individual bacterial taxon in lung cancer evolution (51).

Cigarettes may modify commensal microbiome by microbes in tobacco, tobacco-induced defect of antimicrobial defenses, increased biofilm formation by certain advantageous bacteria, or altered local microenvironment (e.g., oxygen tension) (52). Yu et al. indicated that chronic exposure to tobacco smoke was associated with higher diversity of lung microbiota and the higher proportion of smokers in LUSC may be the explanation (53). Long-term inhalation of tobacco-related detrimental particles could increase the airborne spread of microbes and impede the elimination of microbes from the respiratory system (53). As a result, increased microbial burden might be found in smokers. However, Greathouse et al. and our group found no significant changes in alpha-diversity by smoking status. For LUSC patients, Greathouse et al. showed that the relative abundance of Acidovorax was higher in former and current smokers (24). Zheng et al. reported that increased Pseudoalteromonas sp. CF149, Roseburia hominis, Penicillium expansum, etc. and decreased Pseudomonas mosselii and Pseudomonas putida were shown in mixed samples of bronchoscopy and lobectomy in NSCLC patients with a smoking history (19). The following reasons may help elucidate those differences of taxonomic relative abundance with different tobacco exposure. Firstly, previous studies included both LUSC and LUAD in the analyses, and the greater proportion of smokers in the LUSC group may explain the differences. LUAD mainly arises from peripheral airways, and its association with smoking is weaker compared with LUSC. As a result, the effect of cigarettes on the respiratory microbiome of LUAD may be alleviated. Secondly, increased microbial burden in smokers does not necessarily mean higher diversity. Being continuously exposed to the external environment, the respiratory microbiome undergoes dynamic changes. The immigrant microorganisms could become part of the resident microbiota, but the majority belongs to the transient group, which could not grow in the respiratory tract and contribute little to respiratory microbial diversity (48). Moreover, different sample types and processing procedures might also contribute to the variations. The exact causal relationship between cigarettes and microbiome alteration in lung cancer patients remains to be further explored (52).

There are several potential limitations in the current study. First, it is difficult to balance baseline characteristics among different groups due to the disease nature and consecutively enrolled cohort. As a result, the inter-group age, gender, smoking status, degree of differentiation, and TNM stage did not match. Alterations of airway microbiome with respect to those characteristics were not evaluated. Second, the limited sample size, especially the BPD group, might compromise the power of this study. In the comparison of taxonomic abundance among different pathological stages of lung cancer, no bacterial taxa showed significant difference. Whether microbial composition differs among lung cancer stages remains to be explored. Studies with sufficient and evenly distributed samples are warranted to further validate the current results. Third, this study was cross-sectional and unable to determine the temporal relationship of the respiratory microbiome to onset or progression of lung cancer. Fourth, 16S rRNA amplicon sequencing is liable to contamination during PCR procedures and generally classifies dominant bacteria to the genus level. Shotgun metagenomic sequencing may help make up these defects and identify more specific species within these genera for each lung sample (54). Fifth, OTU-level flow, Mothur, was still used in our study to be comparable to studies about respiratory microbiota in lung cancer patients. ASV (amplicon sequence variant)-level pipelines (DADA2, QIIME2-Deblur, and USEARCH-UNOISE3) are justified to confirm the current results. In addition, 16S rRNA sequencing cannot provide information on the viability of bacteria and bacteria–host interactions. Bacterial culture, respiratory microbiota transplantation, and metabonomics are potential ways to clarify the role of flora activities in NSCLC.

In conclusion, we found significantly different respiratory microbiome taxa, abundance, and diversity in lung cancer of different pathology and some stages. Short-term antibiotic application might play a minor role in molding airway microbiota in lung cancer patients. Composition and diversity of respiratory microbiota in lung adenocarcinoma are not affected by cigarette exposure. Prospective longitudinal studies that utilize bacterial culture, respiratory microbiota transplantation, and metabonomics are needed to further clarify the role of respiratory microbiota activities in NSCLC.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: BioProject, accession number PRJNA834019.

The studies involving human participants were reviewed and approved by the Institutional Review Board committee of West China Hospital, Sichuan University. The patients/participants provided their written informed consent to participate in this study.

Study conceptualization and design: YH, XZ. Supervision and administrative support: YH. Funding acquisition: XZ. Specimen processing and data collection: XZ, XL. Data analysis and interpretation: all authors. Manuscript writing and editing: all authors. All authors contributed to the article and approved the submitted version.