This retrospective study explored the use of mNGS in diagnosing sputum samples from COVID-19 patients, aiming to identify pathogens and examine respiratory flora distribution and microecological shifts. Although often overlooked, respiratory and gastrointestinal flora changes are significant in disease etiology, progression and prognosis38,39,40,41,42,43. Understanding SARS-CoV-2’s interaction with respiratory flora is paramount, and mNGS provides a comprehensive microbial landscape for accurate diagnosis and informed therapy.
In recent studies, sputum, bronchoalveolar lavage fluid, and nasal swabs have been utilized for the diagnosis of COVID-19 infection44 However, limited research exists on the alterations of the respiratory tract microbiota and the potential distribution patterns of pathogens among COVID-19 patients. Additionally, while mNGS has been extensively employed in the investigation of gastrointestinal infections45bloodstream infections23,46,47 and pulmonary infections23,25,48 there is a relative paucity of studies focusing on lower respiratory tract infections primarily caused by SARS-COV-2. In this comprehensive study, we employed mNGS for the first time to elucidate the distribution of respiratory pathogens and the shifts in respiratory tract microbiota among patients with COVID-19.
Routine culture guides clinical decisions but depends on institutional expertise and typically identifies only dominant pathogens, missing slow-growing or non-dominant ones. mNGS offers rapid, unbiased nucleic acid detection for viruses, bacteria, fungi, and other pathogens without culture. Its strength lies in diagnosing complex diseases and identifying unknown pathogens.
Previous research has highlighted the challenge of mNGS’s relatively high false positive rate, attributed to the presence of human information in the samples that is challenging to eliminate49. Despite this, limited studies have been conducted to assess the concordance between mNGS and conventional detection methods. Understanding the factors that contribute to these false positives and exploring strategies to mitigate them is crucial for enhancing the reliability and accuracy of mNGS in clinical settings.
In this comprehensive study, we delved into the concordance between the mNGS analysis of sputum samples and the traditional routine culture methods. Our findings revealed that the level of agreement between mNGS and conventional culture results was somewhat lower than expected, particularly when compared to clinical outcomes. This discrepancy might be attributed to the insufficient microbial reads obtained from the samples.In this study, the Bacteria, fungi, virus detected by mNGS analysis needs to be tested with clinical symptoms, x-ray evidence, traditional microbiological examination, mNGS, serological examination and other tests to determine whether it is a pathogen.
The results of our study demonstrated a notably high positive rate of 95.35% for mNGS. In terms of pathogen coverage, mNGS accounted for 36.36% (12/33) of the total bacterial isolates obtained through conventional culture and an impressive 74.07% (20/27) of the fungal isolates. These findings underscore the potential of mNGS in detecting non-dominant pathogenic bacteria that might be overlooked by conventional culture methods. Furthermore, mNGS has the added advantage of identifying rare pathogens and mitigating the interference caused by antibiotics.
While there is ample room for improvement in terms of mNGS’s concordance with conventional methods, its unique capabilities position it as a promising tool in clinical settings. Its ability to provide a comprehensive analysis of the microbial composition within a sample, including both dominant and non-dominant pathogens, offers invaluable insights for clinicians in treatment decision-making and prognosis assessment20.
It is crucial for individuals suffering from fungal infections to achieve prompt diagnosis of fungal pulmonary infection. Recently, there has been a steady rise in the incidence of fungal infections, yet diagnosing early pulmonary fungal infection remains challenging. The current clinical approach to fungal diagnosis primarily relies on the G test and GM test. However, the diagnostic accuracy for early pulmonary infection is not entirely satisfactory, often requiring multiple tests, which can not only exacerbate the patient’s discomfort but also delay appropriate treatment. Given its promising performance, mNGS is anticipated to play a pivotal role in the early detection of fungal infections.
In a recent study, we observed that mNGS exhibits promising results in the diagnosis of fungal infections, enhancing sensitivity in detecting fungal infections. Notably, there were discrepancies in fungal culture results compared to mNGS findings in 10 cases, as detailed in Table 4. Therefore, it is anticipated that the combined utilization of clinical detection methods such as the G test, GM test, or pathological examination will enhance the detection rate of fungal infections.
It is imperative to emphasize that accurate and timely diagnosis of fungal infections is crucial for effective treatment and patient outcomes. The integration of advanced diagnostic tools like mNGS with traditional methods offers a promising approach to addressing this challenge. Future research should further explore the potential of these combined diagnostic strategies to improve the detection and management of fungal infections.It is noteworthy that while specific pathogens and clinical indicators showed associations with outcomes, the overall microbial burden (total number of species or reads detected by mNGS) did not demonstrate a significant independent correlation with prognosis in our analysis.
SARS-COV-2 poses a significant threat to human life and health, with an intricate pathogenesis that often affects multiple organs and systems, primarily targeting the respiratory system7,50. The pathogenesis of SARS-CoV-2 is intricate, often affecting multiple organs and systems, with the respiratory system being the primary target of infection. Beyond its direct cytopathic effects, emerging hypotheses suggest that SARS-CoV-2 may exhibit bacteriophage-like behavior, potentially directly impacting the respiratory microbiota composition51. This interaction could disrupt the delicate ecological balance of the respiratory tract52diminishing the abundance and diversity of commensal bacteria that are essential for maintaining immune homeostasis53 and for providing colonization resistance against pathogens54. Our findings, which showed a significantly reduced respiratory microbiota abundance in critically ill patients, align with this paradigm. The depletion of this protective microbial layer may create a permissive environment for the proliferation of opportunistic pathogens55thereby exacerbating the disease course. This underscores the importance of a healthy, resilient respiratory flora as a key modulator of host defense against SARS-CoV-2, and its disruption represents a crucial aspect of the virus’s pathogenesis.The detection of pathogens by mNGS in culture-negative samples underscores the technique’s high sensitivity, yet necessitates cautious interpretation. As established in respiratory medicine, the presence of microbial nucleic acid does not invariably indicate active disease but may represent colonization or a state contained by the host’s immune system and the competitive inhibition of a healthy microbiota55. Furthermore, our data revealed several cases where mNGS detected pathogens that were not identified by standard culture. As previously discussed, this is a well-known phenomenon where the healthy microbiota plays a critical role in providing ’colonization resistance,’ thereby suppressing the expansion and pathogenicity of invading microbes56. The functional profile of a robust microbiome, characterized by the production of bacteriocins, competitive exclusion for nutrients, and maintenance of a favorable microenvironment, is essential for this protective effect. Our findings align with this concept, suggesting that mNGS can provide a more comprehensive view of the microbial landscape, including pathogens that are present but held in check by the host’s microbial community. Therefore, the clinical relevance of mNGS findings must be rigorously correlated with the patient’s symptoms, radiological evidence, and inflammatory markers to accurately distinguish between infection and colonization-a process central to our interpretation strategy. This approach is particularly crucial given that previous studies have demonstrated SARS-CoV-2 infection induces alterations in the abundance and distribution of the respiratory tract microbiota, while conventional detection methods like microbial culture and targeted nucleic acid tests have limitations in comprehensively capturing these complex microbial dynamics in patients57.
Recently, the existence of the lower respiratory tract microbiota has been clarified. Advances in genetic analysis have enabled the analysis of the lower respiratory tract microbiome through macrogenomic techniques. These advancements hold the potential to contribute significantly to personalized medicine, offering new biomarkers for diagnosis and treatment58.
In this study, it was found that the abundance of respiratory tract microflora in critically ill patients with COVID-19 was significantly lower than that in COVID-19 (Non-critically ill patients) and control groups without COVID-19 infection, which may be due to the fact that SARS-COV-2 infection destroyed the distribution of respiratory tract colonization flora. In this paper, the changes of respiratory tract flora among different groups of COVID-19 were studied by mNGS for the first time, and the characteristics of respiratory tract flora of COVID-19 patients were revealed. At present, many studies have proved that the changes of respiratory tract flora are closely related to the occurrence, development and prognosis of the disease. The change trend of respiratory tract flora in patients with COVID-19 may become a new idea for the diagnosis and treatment of COVID-1959.
We included mNGS-derived metrics (such as the number of microbial sequences and types of pathogens detected) alongside clinical indicators in the analysis of patient prognosis. While the total number of microbial species or sequencing reads was not independently associated with clinical outcomes in our multivariate model, the detection of specific pathogens (e.g., Enterococcus faecium) was associated with critical illness. Furthermore, clinical inflammatory markers (CRP, PCT, NLR, etc.) remained the strongest correlates of poor prognosis. This indicates that the clinical utility of mNGS in prognostic prediction lies not in the sheer quantity of detected microbes, but in the identification of high-risk pathogens which, when integrated with host inflammatory response data, can provide a more comprehensive prognostic picture. Therefore, the mNGS findings are not prognostically irrelevant but must be interpreted in a nuanced clinical context.
The present investigation, while providing valuable insights, must be viewed with certain limitations. It is important to note that our machine learning analysis for predicting critical illness is exploratory in nature, limited by the modest sample size. While the combination of inflammatory markers and specific respiratory pathogens appears promising for severity stratification, the risk of overfitting is high. Thus, these results are primarily hypothesis-generating and underscore the need for future prospective studies with dedicated external validation cohorts to confirm the clinical utility of such a model. Secondly, the small sample size utilized in this study (as mentioned above in the context of the machine learning analysis) further limits its generalizability and the ability to draw definitive conclusions from the overall dataset. Therefore, future research endeavors should prioritize larger sample sizes to further elucidate the utility of mNGS in COVID-19 and LRTI. Additionally, this study does not capture the unique applications of mNGS in other infectious diseases, especially in challenging cases and the detection of previously unknown pathogens. Future studies should aim to address these gaps in knowledge. In fact, the reported results of mNGS are oriented towards the suspected pathogenic flora, and the normal microbiota is often rarely presented in the reports of mNGS. In the same type of research60,61,62this defect is inevitable, which may be an important development direction for mNGS in the future. We hope that mNGS can better present normal microbiota and pathogens through technological progress in the future.
To summarize, mNGS exhibits remarkable proficiency in rapidly detecting pathogens within LRTI, surpassing the positive rate of conventional culture methods. Furthermore, among critically ill patients with COVID-19, a significant reduction in the abundance of respiratory tract microflora was observed, along with statistically distinct patterns in the distribution of pathogens.
