Performance of metagenomic next-generation sequencing, Xpert MTB/RIF and acid-fast staining for diagnosing tuberculous pleurisy and empyema

Introduction

Tuberculosis (TB) is mainly caused by Mycobacterium tuberculosis (MTB), which is also the number one cause of death from a single infectious source and the 13th cause of death in the world. Fortunately, the diagnosis and treatment of tuberculosis have been widely recognized and valued worldwide, and the relevant diagnosis and treatment standards are gradually improving. There are still shortcomings in various detection methods for tuberculosis. MTB culture and acid-fast staining (AFS) are currently important methods for clinical diagnosis of MTB. The sensitivity of culture in diagnosing MTB is low, and it takes a long time and often delaying the diagnosis and treatment of the patient’s condition. Its clinical application value in early diagnosis of MTB is limited (1). Therefore, other corresponding detection methods are flourishing, including molecular detection, antibody detection, etc. Xpert MTB/RIF (Cepheid, USA), based on real-time polymerase chain reaction (PCR), is one of the commonly used molecular diagnostic products, which has been confirmed to have excellent diagnostic performance by multiple studies (2-4). Denkinger CM et al. conducted a meta-analysis on the performance of Xpert MTB/RIF in diagnosing extrapulmonary tuberculosis (5). This analysis included 18 studies with a total of 4,461 samples, and the results showed that the sensitivity of Xpert differed substantially between sample types. In lymph node tissues or aspirates, Xpert pooled sensitivity was 83.1% [95% confidence interval (CI): 71.4–90.7%] versus culture and 81.2% (95% CI: 72.4–87.7%) versus composite reference standard (CRS). In cerebrospinal fluid, Xpert pooled sensitivity was 80.5% (95% CI: 59.0–92.2%) against culture and 62.8% (95% CI: 47.7–75.8%) against CRS. In pleural fluid, pooled sensitivity was 46.4% (95% CI: 26.3–67.8%) against culture and 21.4% (95% CI: 8.8–33.9%) against CRS. Xpert pooled specificity was consistently >98.7% against CRS across different sample types.

Metagenomic next-generation sequencing (mNGS) is increasingly used in clinical applications, and its application in the diagnosis of infectious diseases is also increasingly attracting scientific research and clinical attention. Due to the advantages of mNGS, such as short time, unbiased detection, and complete collection of species, it is gradually changing the traditional way of diagnosis of infectious diseases, and it can conduct unbiased sequencing of microorganisms in human tissue samples and body fluid samples, including blood, cerebrospinal fluid, urine, and sputum (6). Study showed that the sensitivity of mNGS in diagnosing suspected MTB was 58.8% (20/34) with 100% (16/16) specificity, while in confirmed patients, the sensitivity of mNGS diagnosis was 63.6% (14/22), and the sensitivity of traditional detection methods combined with mNGS for diagnosing MTB was as high as 82.4% (28/34) (7). Another study that included 208 suspected patients with negative staining of MTB smears from extrapulmonary specimens showed that 45 cases tested positive for mNGS. The positivity rate of mNGS detection for MTB was the highest among various types of extrapulmonary tuberculosis; the area under the curve (AUC) value was the highest among various detection methods (0.79). Among them, 33 patients tested positive for mNGS, while other tests (including bacterial culture and Xpert MTB/RIF technology) tested negative (8). The above research indicates that mNGS has great value in the diagnosis of tuberculosis.

In this study, fresh tissues, pleural fluid or pus were collected from suspected tuberculosis patients. We tested these samples using mNGS. At the same time, some samples were subjected to AFS, Xpert, and other tests. We compared the effectiveness of different detection methods to evaluate their diagnostic ability for MTB. We present this article in accordance with the STARD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1411/rc).

Methods

Clinical specimens

All samples in this study were collected from the Department of Thoracic Surgery at the Second Hospital of Nanjing, China. This study was approved by the Ethics Committee of Nanjing Hospital Affiliated to Nanjing University of Chinese Medicine (No. 2024-LS-ky-020). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Each enrolled patient or their family signed a written informed consent. From January 2021 to January 2023, the patients who were suspected or required isolation for tuberculosis were included in the study. After excluding factors such as inability to collect samples and obtain informed consent, 154 patients were successfully enrolled. The types of samples collected include fresh tissue, pleural fluid or pus. All collected samples were subjected to mNGS, with some samples undergoing AFS and tuberculosis Xpert testing. A few samples undergo other tests, such as T-cell Spot Test (T-SPOT) and culture for clinical diagnosis. The reasons why some patients did not undergo comprehensive testing are: (I) the sample volume is too small to conduct all tests; (II) the results of partial testing and other examination methods are sufficient for doctors to make a clear diagnosis; (III) the patient is unwilling to undergo further testing due to cost and other reasons. Sequence data that support the findings of this study have been deposited in NCBI with the primary accession code PRJNA1195977.

mNGS detection and bioinformatics analysis

Sample collection and nucleic acid extraction

All the samples were collected during the surgery or lung biopsy for AFS, mNGS, Xpert or other examinations. All the samples were detected by mNGS. The whole process was performed by Genoxor Medical Science and Technology Inc. (Shanghai, China). After pre-treatment, the nucleic acid extraction was performed using the TIANamp Micro DNA Kit (Tiangen Biotech, Beijing, China) or Biomarker Micro Cell/Tissue Total RNA Isolation Kit (Biomarker Technologies, Beijing, China). Quantitative analysis of nucleic acid using Qubit 3.0 Fluorometer and Quant-iT dsDNA HS Assay Kit (ThermoFisher Scientific, Massachusetts, USA).

mNGS sequencing

The extracted nucleic acid is fragmented using transposase. Nextera XT DNA Library Preparation Kit (Illumina, California, USA) was used for libraries construction. Library construction includes end pair, adapter alignment, and 5–7 PCR cycles, resulting in a final library of approximately 1 µg. The libraries was analyzed by an Agilent 2100 Bioanalyzer (Agilent Technology, California, USA). Metagenome shotgun sequencing was performed in a single-end 75bp mode on the NextSeqTM 550DX sequencer, using the NextSeq 500/550 High Output Kit. No template control (NTC) samples were sequenced simultaneously to monitor pollution.

Bioinformatics analysis

For trimming adapter sequences and removing low-quality tails and reads, a quality control process for raw data were first performed. Then, the clean reads were filtered by aligning to the human reference genome GRCh37 with the short-read alignment tool Bowtie v2.2.6 (9), with the mapped reads excluded. The identically duplicated reads were considered as technical artifacts by PCR (10) and were de-duplicated. The reads number in the Kraken classification report was estimated by the Bayesian algorithm implemented by Bracken v2.2. The microbial taxa bearing putative contamination introduced during the experiments were removed based on the information of NTC samples. The estimates of the percentage relative abundance of each species were computed for bacteria, viruses (excluding phages), and fungi, respectively, in a similar way to TPM (transcripts per million) described previously (11).

AFS

AFS is completed by the laboratory department of the Second Hospital of Nanjing.

• First, add a drop of physiological saline on the glass slide, and then evenly spread the specimen in the physiological saline to make a thin and uniform smear. The size of the smear is generally around 1 cm × 2 cm. Let the smear dry naturally, and fix the dried smear 3–4 times with a flame.
• Cover the smear with carbolic acid fuchsin staining solution. Place the smear on the staining rack and gently heat it near the flame of an alcohol lamp for 5–10 minutes. After staining is complete, let it cool naturally and then slowly rinse the smear with running water.
• Use a 3% hydrochloric acid alcohol solution to decolorize for 30 seconds to 1 minute until the red color on the smear no longer fades. Immediately rinse the smear with running water to terminate the decolorization reaction.
• Perform counterstaining with methylene blue dye or Lü’s alkaline methylene blue dye for 2–3 minutes. Rinse the smear with running water to remove excess dye, and then let the smear dry naturally.

Xpert testing

Xpert MTB/RIF^® kits were purchased from Cepheid, USA. Detection is completed by the laboratory department of the Second Hospital of Nanjing according to the instructions.

Diagnostic criteria for tuberculosis

The final clinical diagnosis of all patients is jointly reviewed by a multidisciplinary team (MDT). The team is composed of thoracic surgeons, tuberculosis physicians, radiologists, and pathologists, who jointly evaluate all test results and clinical data. If any of the following 2–4 criteria are met, the sample is considered positive for tuberculosis:

• Clinical manifestations: persistent cough for more than 2 weeks, accompanied by symptoms of tuberculosis poisoning such as hemoptysis, low-grade fever, night sweats, and weight loss, and ineffective treatment with antibiotics.
• Pathogenic examination.
• Positive AFS of sputum smear (at least 2 samples).
• MTB culture is positive.
• Molecular testing (such as Xpert MTB/RIF) is positive.
• Imaging features: chest X-rays or computed tomography (CT) scans show typical changes such as upper lung field infiltration, cavity formation, fibronodular lesions, and mediastinal lymph node enlargement.
• Pathological evidence: tissue biopsy revealed caseous necrotic granulomas or acid-fast bacilli.
• Immunological testing: the gamma interferon release test (IGRA) or tuberculin skin test (TST) is positive.

Statistical analysis

The CIs of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated on the website: www.vassarstats.net. Significance of sensitivity was calculated on the module “Significance of the Difference Between Two Independent Proportions”.

Results

Clinical characteristics of patients in this study

In this study, a total of 154 suspected or requiring exclusion of tuberculosis patients were successfully enrolled, with one sample per patient. All 154 samples underwent mNGS testing, 99 underwent tuberculosis Xpert testing, and 143 underwent AFS. A small portion of the samples have undergone culture, T-SPOT testing or other tests. Clinicians make a comprehensive diagnosis and develop treatment plans for patients based on the above results and other examinations, as shown in Figure 1.

Figure 1 Study design. mNGS, metagenomic next-generation sequencing; MTB, Mycobacterium tuberculosis; RIF, rifampicin; Xpert, Xpert MTB/RIF.

Among all the patients in this study, 119 were male and 35 were female. The patients have a wide age distribution, with an average age of 43.36 years, ranging from 6 to 77 years, and a median age of 43 years. The average body mass index (BMI) of patients is 21.99 kg/m², ranging from 14.34 to 31.96 kg/m², with a median of 22.05 kg/m². According to the patients’ statement, most of them had no chronic disease history (n=101, 65.58%), and the three most chronic disease histories were tuberculosis (n=16, 10.39%), hypertension (n=17, 11.04%), and diabetes (n=7, 4.55%). All of these patients underwent surgical treatment afterwards. The average white blood cell (WBC) value is 6.46×10⁹/L, ranging from 2.12×10⁹/L to 24.55×10⁹/L, with a median of 5.78×10⁹/L. The average neutrophil (NE) value is 4.35×10⁹/L, ranging from 0.02×10⁹/L to 15.52×10⁹/L, with a median of 3.66×10⁹/L, as shown in Table 1. Except for one patient (P35) who was automatically discharged, all others improved and were discharged from the hospital. Information and test results of all patients are shown in Table S1.

The performance of three detection methods in tuberculosis diagnosis

Firstly, we calculated the positivity rate of mNGS, Xpert, and AFS methods for detecting tuberculosis in all samples. The comprehensive positivity rate of mNGS was 48.05% (74/154), Xpert was 44.44% (44/99), and AFS was 57.34% (82/143). We further conducted separate positivity rate analysis on different sample types. There were two sample types in this study: 114 fresh tissues and 40 pleural fluid or pus. For fresh tissues, the results showed that the positivity rate of AFS (60/107, 56.07%) was slightly higher than that of mNGS (59/114, 51.75%), while Xpert had the lowest positivity rate (37/79, 46.84%). The positivity rate of AFS of pleural fluid or pus (22/36, 61.11%) was higher than that of mNGS (15/40, 37.50%) and Xpert (7/20, 35.00%). The results are shown in Table 2. These results indicate that the AFS method has the highest positivity rate, mainly reflected in the types of pleural or pus samples. As a molecular diagnostic method, the positivity rate of mNGS is higher than Xpert, mainly reflected in fresh tissue sample types.

Next, we calculated and compared the sensitivity, specificity, PPV and NPV of the three methods. The results showed that the sensitivity, specificity, PPV and NPV of mNGS were 55.22%, 100%, 100% and 25%, respectively. The sensitivity, specificity, PPV and NPV of Xpert were 48.35%, 100%, 100%, 18.97%, respectively. The sensitivity, specificity, PPV and NPV of acid-fast stain were 58.59%, 53.33%, 91.46%, 13.11%, respectively. The results of difference analysis showed that there was no significant difference in sensitivity among the three methods, as shown in Table 3.

Comparison of three detection methods in the same sample

Considering that some samples did not undergo Xpert and AFS tests, we selected samples that underwent both mNGS, Xpert, and AFS tests and directly compared their detection performance. A total of 93 samples underwent three types of testing. According to each method, the test results are positive or negative, and there are a total of 8 combinations of results. There are two types of results that are completely consistent: mNGS+/Xpert+/AFS+ and mNGS−/Xpert−/AFS−, with a total proportion of 45.16% (42/93). There are six types of inconsistent results: mNGS+/Xpert+/AFS−, mNGS−/Xpert−/AFS+, mNGS+/Xpert−/AFS+, mNGS−/Xpert+/AFS−, mNGS+/Xpert−/AFS−, mNGS−/Xpert+/AFS+, with a total proportion of 54.84% (51/93), as shown in Figure 2.

Figure 2 The distribution of consistency in the results of three detection methods. The circles marked with mNGS, Xpert, or AFS represent all positive results of this detection method. The ‘+’ in the comments within the circle represents a positive result, while the ‘−’ represents a negative result; ‘n’ represents the number of samples, followed by the proportion of the represented portion of samples to all samples. AFS, acid-fast staining; mNGS, metagenomic next-generation sequencing; MTB, Mycobacterium tuberculosis; RIF, rifampicin; Xpert, Xpert MTB/RIF.

For these 93 samples, we also analyzed the sensitivity, specificity, PPV and NPV. The results showed that the sensitivity, specificity, PPV and NPV of mNGS were 65.88%, 100%, 100%, 21.62%, respectively. The sensitivity, specificity, PPV and NPV of Xpert were 49.41%, 100%, 100%, 15.69%, respectively. The sensitivity, specificity, PPV and NPV of acid-fast stain were 57.65%, 62.50%, 94.23%, 12.20%, respectively. The differential analysis results show that there is a significant difference in sensitivity between mNGS and Xpert, as shown in Table 4.

Discussion

In this study, we used mNGS, Xpert, and AFS methods to test clinical samples of suspected tuberculosis patients, and compared the results to analyze their application value in diagnosis for tuberculosis. The results showed that the sensitivity of mNGS was higher than that of Xpert, and the specificity of mNGS and Xpert was excellent, both higher than AFS.

mNGS, as a highly sensitive and specific molecular detection technology, has gradually been applied in clinical diagnosis of infectious pathogens. Multiple research results have shown that mNGS has great potential in clinical molecular diagnosis (12-15). In this study, mNGS performed better than Xpert in the diagnosis of tuberculosis, indicating that mNGS has important value in the diagnosis of tuberculosis. Previous studies have shown that mNGS is not inferior to Xpert in diagnostic performance (16-18). The performance of AFS in this study was better than the other two methods for specific samples.

The type of sample usually affects the performance of molecular detection, and tuberculosis infection may affect various parts of the body, such as tuberculous meningitis, pleurisy, etc., resulting in a diversity of tested samples. Multiple studies have shown that the same detection method performs differently in different sample types, and this is also true for mNGS (19,20). Previous studies on the application of mNGS in tuberculosis diagnosis have mainly focused on bronchoalveolar lavage fluid (BALF). The sample types used in this study are surgical tissue, pleural fluid or pus. Currently, research on the application of mNGS in tuberculosis diagnosis is rarely focused on these sample types. Our results indicate that mNGS performs better than Xpert in terms of tissue, pleural fluid or pus. The performance of mNGS in other sample types requires further research and exploration, and mNGS may be the best molecular detection method in some specific sample types.

There are also some limitations in this study. Firstly, in terms of patient enrollment, as this study is a single-center study, all enrolled patients require thoracic surgery. In addition, some patients have a history of tuberculosis, and the diagnosis of these patients has been basically clear. Secondly, as this study is a retrospective study, some patients did not undergo Xpert or AFS testing for various reasons. This leads to differences between the analysis based on all patients and the analysis based on patients undergoing all three detection methods. Thirdly, this study did not establish a reference method, and the final clinical diagnosis made by clinical doctors may be non-objective. To avoid this issue as much as possible, the final clinical diagnosis of all patients underwent multidisciplinary consultation.

Due to the characteristics of technical principles, various detection methods still have their own advantages. As a quantitative indicator of microorganisms in mNGS results, the reads number has been proven to be related to various factors (21,22). Liu et al. found a positive correlation between the reads number of herpesvirus in mNGS and patient mortality in severe pneumonia patients. The author believes that a high reads number of herpesvirus may indicate a weakened immune system in patients, and therefore can predict a poor prognosis (22). In this study, we found that among the mNGS positive results, the reads number with completely consistent results from the three methods was higher than those with inconsistent results (average reads number 2,513 vs. 131, P=0.268). This indicates that reads number may become an important indicator for positive interpretation of mNGS results. In addition to detecting tuberculosis, mNGS can also detect other microorganisms in samples, making it suitable for the diagnosis of tuberculosis mixed with other pathogenic infections, as well as the differential diagnosis of non-tuberculosis infections. The mNGS results in this study revealed bacterial and fungal infections in some patients. In addition, for samples with complex microbiota, such as BALF and sputum derived from the respiratory tract, mNGS can display their microbiota profiles. Although this aspect has not yet been applied in clinical practice, relevant research has been conducted (23). Although mNGS has excellent detection performance, its reliance on large detection machines, strict laboratory facilities, and professional testing personnel currently limits its large-scale application. In China, the cost of mNGS testing is about $500 per sample and the testing time is about 24 hours, which is several times higher than the other two testing methods. Due to its high enrichment of target gene regions in detection samples, Xpert has advantages in identifying drug-resistance genes with significant heterogeneity in abundance, making it more suitable for drug resistance monitoring than mNGS. AFS has more advantages in terms of detection convenience, detection time, and economic applicability. Due to differences in personnel, equipment, and medical policies around the world, it is usually necessary to screen for diagnostic combinations that are suitable for the local area in clinical practice. The comparison results of three methods for 93 samples in this study provide a reference for clinical physicians. The application of mNGS in clinical infection diagnosis is still in its preliminary stage, and more research is needed in the future to explore its application value.

Conclusions

Our research results indicate that mNGS and AFS have higher sensitivity compared to Xpert, while mNGS and Xpert have higher specificity. Based on the above results, the combination of mNGS and AFS can be considered for preliminary diagnosis of tuberculosis in certain scenarios, such as as no need to consider the cost or when the patient is critically ill.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1411/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1411/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1411/prf

Funding: This study was supported by the Social Development Project of the Key Research and Development Plan of Jiangsu Province (No. BE2023660); Innovation Center for Infectious Disease of Jiangsu Province (No. CXZX202232), and “333 Talent Project” of Jiangsu Province.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1411/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by the Ethics Committee of Nanjing Hospital Affiliated to Nanjing University of Chinese Medicine (No. 2024-LS-ky-020). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Each enrolled patient or their family signed a written informed consent.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bahr NC, Boulware DR. Methods of rapid diagnosis for the etiology of meningitis in adults. Biomark Med 2014;8:1085-103. [Crossref] [PubMed]
2. Steingart KR, Schiller I, Horne DJ, et al. Xpert® MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults. Cochrane Database Syst Rev 2014;2014:CD009593. [Crossref] [PubMed]
3. Kohli M, Schiller I, Dendukuri N, et al. Xpert MTB/RIF Ultra and Xpert MTB/RIF assays for extrapulmonary tuberculosis and rifampicin resistance in adults. Cochrane Database Syst Rev 2021;1:CD012768. [Crossref] [PubMed]
4. Vergara Gómez A, González-Martín J, García-Basteiro AL. Xpert® MTB/RIF: Usefulness for the diagnosis of tuberculosis and resistance to rifampicin. Med Clin (Barc) 2017;149:399-405. [Crossref] [PubMed]
5. Denkinger CM, Schumacher SG, Boehme CC, et al. Xpert MTB/RIF assay for the diagnosis of extrapulmonary tuberculosis: a systematic review and meta-analysis. Eur Respir J 2014;44:435-46. [Crossref] [PubMed]
6. Chiu CY, Miller SA. Clinical metagenomics. Nat Rev Genet 2019;20:341-55. [Crossref] [PubMed]
7. Lin A, Cheng B, Han X, et al. Value of next-generation sequencing in early diagnosis of patients with tuberculous meningitis. J Neurol Sci 2021;422:117310. [Crossref] [PubMed]
8. Sun W, Lu Z, Yan L. Clinical efficacy of metagenomic next-generation sequencing for rapid detection of Mycobacterium tuberculosis in smear-negative extrapulmonary specimens in a high tuberculosis burden area. Int J Infect Dis 2021;103:91-6. [Crossref] [PubMed]
9. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods 2012;9:357-9. [Crossref] [PubMed]
10. Martin TC, Visconti A, Spector TD, et al. Conducting metagenomic studies in microbiology and clinical research. Appl Microbiol Biotechnol 2018;102:8629-46. [Crossref] [PubMed]
11. Li B, Ruotti V, Stewart RM, et al. RNA-Seq gene expression estimation with read mapping uncertainty. Bioinformatics 2010;26:493-500. [Crossref] [PubMed]
12. Liu J, Dong Y, Huang Y, et al. Clinical metagenomic next-generation sequencing test for diagnosis of central nervous system infections in ICU: a multicenter retrospective study. Int J Infect Dis 2025; Epub ahead of print. [Crossref]
13. Chen J, Lou H, Yao Y, et al. Metagenomic next-generation sequencing for concurrent diagnosis of infections and malignancies in patients with lung lesions: a single-center prospective study. Sci Rep 2025;15:33721. [Crossref] [PubMed]
14. Han D, Li Z, Li R, et al. mNGS in clinical microbiology laboratories: on the road to maturity. Crit Rev Microbiol 2019;45:668-85. [Crossref] [PubMed]
15. Liu M, Cheng Y, Wang X, et al. Diagnosing Rickettsia felis infection with Metagenomic Next-Generation Sequencing (mNGS) in a patient with ankylosing spondylitis: a case report and literature review. Front Immunol 2025;16:1643599. [Crossref] [PubMed]
16. Yan L, Sun W, Lu Z, et al. Metagenomic Next-Generation Sequencing (mNGS) in cerebrospinal fluid for rapid diagnosis of Tuberculosis meningitis in HIV-negative population. Int J Infect Dis 2020;96:270-5. [Crossref] [PubMed]
17. Sun WW, Gu J, Fan L. Application value of metagenomic next-generation sequencing (mNGS) in the diagnosis of different types of tuberculosis. Zhonghua Jie He He Hu Xi Za Zhi 2021;44:96-100. [Crossref] [PubMed]
18. Li Y, Jiao M, Liu Y, et al. Application of Metagenomic Next-Generation Sequencing in Mycobacterium tuberculosis Infection. Front Med (Lausanne) 2022;9:802719. [Crossref] [PubMed]
19. Miao Q, Liang T, Pei N, et al. Evaluation of respiratory samples in etiology diagnosis and microbiome characterization by metagenomic sequencing. Respir Res 2022;23:345. [Crossref] [PubMed]
20. Li X, Liang S, Zhang D, et al. The clinical application of metagenomic next-generation sequencing in sepsis of immunocompromised patients. Front Cell Infect Microbiol 2023;13:1170687. [Crossref] [PubMed]
21. Zhang X, Chen H, Lin Y, et al. Diagnosis of Non-Tuberculous Mycobacterial Pulmonary Disease by Metagenomic Next-Generation Sequencing on Bronchoalveolar Lavage Fluid. Infect Drug Resist 2023;16:4137-45. [Crossref] [PubMed]
22. Liu Y, Wen Z, Fang Y, et al. Herpesvirus reactivation in respiratory tract is associated with increased mortality of severe pneumonia patients and their respiratory microbiome dysbiosis. Front Cell Infect Microbiol 2023;13:1294142. [Crossref] [PubMed]
23. Liu H, Ji S, Fang Y, et al. Microbiome Alteration in Lung Tissues of Tuberculosis Patients Revealed by Metagenomic Next-Generation Sequencing and Immune-Related Transcriptional Profile Identified by Transcriptome Sequencing. ACS Infect Dis 2023;9:2572-82. [Crossref] [PubMed]