This is a single-center retrospective study to assess the diagnostic performance of BALF and lung biopsy tissue for mNGS detecting MTBC. Previous studies have shown the value of mNGS on MTBC detection, most of which generally involved BALF and sputum samples. This is the first study to directly compare BALF with lung biopsy tissue on the use of mNGS in MTBC detection. Preceding studies have indicated that the efficacy of culture, Xpert, and mNGS was significantly affected by anti-TB treatment [18]. All enrolled patients in our research had not previously received anti-TB treatment that ensures the objectiveness and accuracy of the research results. Moreover, we implemented the R-EBUS probe lesions suspected of PTB location when TBLB to make certain of higher diagnostic rates of mNGS from lung biopsy tissues.
The keys to managing infectious diseases include the following vital aspects (1) control the source of infection, (2) cut off the route of transmission, and (3) protect the susceptible population [19]. Early bacteriological diagnosis of PTB case is crucial to limit the spread of infections [20]. Nevertheless, early detection of PTB from other lung infections is difficult by traditional methods, particularly for patients with mild symptoms or uncharacteristic appearances on imaging [21, 22]. In 2020, the incidence and mortality rates of PTB were 47.7644 per 100,000 and 0.1367 per 100,000 in China. The positive tuberculosis case detection rate (CDR) is less than 40% in China, which would place a heavy burden on health care services, and carry enormous consequences for societies and economies [23]. Therefore, it is critical in improving the diagnosis of patients with PTB. Recently, the popularity of mNGS has drawn extensive concerns, particularly with respect to shortening turnaround time and guaranteeing accurate pathogen detection for diagnosis [8]. Xpert is WHO-recommended rapid test that simultaneously detect tuberculosis and rifampicin resistance in people with signs and symptoms of tuberculosis. This assay has been proven useful to diagnose PTB, with 89% sensitivity and 99% specificity [24]. Previous research has shown that Xpert performs similar detection efficiency to mNGS in MTBC detection [18]. In the Health Industry Standards of the People's Republic of China—Tuberculosis Diagnosis issued in 2017 [25], positive Xpert test of MTBC and lung imaging examination were included in the diagnostic standards of tuberculosis, which is highly recognized for the application value of the Xpert test in tuberculosis diagnosis. Removing false-positive results is a primary challenge for mNGS analysis. Therefore, for the patients detected MTBC positive by mNGS, we had implemented Xpert as part of the validation process. In our study, mNGS and Xpert displayed similar overall MTBC detection efficiency, which was also much superior to AFB (Pearson Chi-Square test, < 0.05), which demonstrates that mNGS is a reliable tool for MTBC detection, whether using BALF or lung biopsy specimens. However, during the process of clinical diagnosis and treatment, we noticed that major deficiencies of Xpert are as follows. Firstly, according to the WHO report, it has not improved global case detection rates and shows limited efficacy in extrapulmonary TB (ETB) [26]. Secondly, The Xpert and PCR laboratory tests are DNA tests of MTBC used to diagnose tuberculosis infection, but they cannot diagnose NTM infection [27]. Thirdly, the detection limit of Xpert is 131 CFU/ml; any unit lower than this value cannot be detected, resulting in negative results. Moreover, although Xpert shows both high sensitivity and specificity, and suggests its high value in PTB diagnosis; however, the application of pleural fluid is still limited, and should be improved [28]. Therefore, in some special cases, mNGS can be recognized as an important complement to Xpert in tuberculosis diagnosis. Both methods have their own pros and cons that seem to complement each other. For patients with complex disease, the combining mNGS and Xpert to diagnose PTB can improve the diagnostic rate [29].
Sputum specimens are mostly used as respiratory samples due to advantages such as convenience and security. However, it inevitably exists potential pollution from the oral cavity microbiota and upper respiratory tract (URT). BAL is playing an increasingly important role in the diagnosis of respiratory diseases because it leads in the fields of accountability, practicality and safety for operation [30, 31]. When inflammation lesions are anatomically limited to the surrounding lung parenchyma, the most accessible site should be biopsied. TB, an intracellular bacterium, tends to release fewer extracellular nucleic acids due to its intracellular growth characteristics, which make it difficult to detect [32]. However, the sensitivity for detecting TB will be improved since biopsy may lead to the destruction of tissues and cells. In our research, we found that mNGS samples from BALF and lung biopsy tissue varied in detecting MTBC. Lung biopsy tissue mNGS was more sensitive than BALF to the MTBC detection. Lung biopsy is indeed more invasive in patients than in BALF. Therefore, we should weigh the pros and cons carefully when we can gain the same benefit from both options. For patients with PPLs or who have contraindications of BAL procedures, detection of MTBC using biopsy tissue by mNGS is a more appropriate diagnostic method. Additionally, the sampling method of alveolar lavage fluid is single, while lung biopsy tissue can be obtained by R-EBUS-TBLB, percutaneous lung puncture biopsy (PNLB) and thoracoscopic biopsy. According to clinical observation, most patients have poor tolerance for BAL, particularly for the elderly patients complicated with cardiovascular and cerebrovascular diseases who are rejected for painless tracheoscopy when the risk of anesthesia is high. The pain and discomfort caused by ordinary bronchoscopy is so unbearable that those patients with a specific location of lung lesions prefer to choose PNLB. Besides, thoracoscopic examination for patients with suspected tuberculous pleurisy can accurately determine the lesion site and identify the lesion. If the examination results from thoracoscopic sampling specimens are positive, tuberculous pleurisy can be clearly diagnosed. However, the lung biopsy samples by R-EBUS-TBLB and PNLB are usually small and thus it is not enough to be used for MTBC culture after mNGS and pathological examination. Paraffin-embedded tissue (FFPE) can be used for acid-fast staining identification, but cannot be applied to culture. Tissue-mNGS detection has a unique application value in the abovementioned context. When it comes to the situations that BALF samples have limited value to catch pathogens for special lesion sites (such as PPLs) or the patients have contraindications to BAL procedures, lung biopsy tissue is an optional specimen for MTBC detection by mNGS. MTBC was considered positive when no less than 1 read was mapped to either the species or genus level because of the low yields of DNA extraction and rare specimen contamination. [14, 15]. NGS tends to detect all nucleotide sequences not only from the samples but also those acquired from the contamination, which remains a huge challenge on interpreting of mNGS results. One of the controversial subjects is that the cut-off values for low MTBC sequencing reads by mNGS may indicate a false-positive result. Our study spotted 6 cases with one read mapping while reads of the other 6 cases were less than ten, but these 12 cases were clinically confirmed by successfully responsive to anti-TB treatment. This research revealed that mNGS shows unique advantages and approaches a specificity of 100% in the diagnosis of PTB with a low MTBC burden.
Significantly, although mNGS possesses the ability to identify MTBC and NTM, because of the high internal homology and relatively insufficient coverage, the species within the MTBC was difficult to determine. A total of 11 cases had specific PTB strains among the 36 patients in our research. Based on accumulating evidence, making treatment options tends to depend on different strains. Therefore, it is necessary to explore the future potential of identifying the species MTB. Analogous limitations had also affected the analysis of other pathogens. Species-level are usually not sufficiently precise for characterization as the DNA chains for mNGS DNA sequencing are too short to offer enough information for identification. Therefore, according to the inhomogeneous data on sequencing reads of each pathogen in our study, the pathogens were discussed only at the genera-level while the further research is needed to reveal the inner layer details in species-level of pathogens.
Chronic viral infections can induce inflammatory reactions and suppress processes that drive immune defenses to enhance viral elimination, which increases the severity of MTBC coinfection [33]. MTBC coinfection has been linked to pulmonary Kaposi's sarcoma-associated herpesvirus (KSHV) infections [34]. CMV, EBV and other viruses tend to be detected in the BALF of many patients who are affected by pulmonary infection after kidney transplantation [35]. We found 9 cases complicated with viral infection. Detection of viral DNA or RNA by RT-qPCR remains a gold standard for the identification of infected patients. Unfortunately, independent validation experiments were not carried out in this study owing to this being a retrospective study. Therefore, a diagnosis of co-infection cannot be confirmed. Whether it is necessary to provide antiviral therapy according to a few reads of viruses remains uncertain. However, dynamic variations in the sequencing of reads and relative abundance from mNGS detection before and after treatment could indirectly reflect therapeutic effectiveness and determine whether these viruses were detected as infection or colonization [36]. Hence, more research is needed to affirm the reasonable cut-off values, which are vital elements of therapeutic options.
In our study, the lung virome and mycobiota were analyzed to confirm if they were involved in the alveolar environment. Our study found that the sequence of the microbiota such as Human gammaherpesvirus 4, Human alphaherpesvirus 1, Human betaherpesvirus 5, Human betaherpesvirus 6B, Primate bocaparvovirus 1, Torque teno virus, Torque teno virus 16, Hepatitis B virus, Candida, Aspergillus, Nakaseomyces, and Pneumocystis was detected. We attempted to analyze the relationships and differences among the 27 pathogens detected by mNGS. However, statistical analysis revealed that there was no significant difference among them. Further research is needed to determine the cut-off value of specific pathogens when identifying co-infecting. The detection of some genera that might have a few pathogenic species does not mean that there is a co-infection. More than that, it should confirm the pathogens using complementary methods such as qPCR as mNGS has some challenges in classification and identification. However, mNGS has potential applications in viral identification.
MTBCs can invade the vulnerable groups via the respiratory tract, digestive tract, or broken skin. A study showed that patients with TB had a marked change in bacterial community, then a disappearance of highly homogenous bacterial flora. From our awareness of TB pathology, mycobacteria invasion by injuring the defense of mucosal barrier function in the lower respiratory tract (LRT) is a crucial factor, facilitating the bacterial colonization [37, 38]. However, previous studies merely demonstrated the interaction between MTBC infection and microbiota [39]. The LRT was initially thought to be sterile despite its constant exposure to the environment [40]. However, it has recently been shown to not be a sterile state in the normal LRT with further relevant research [41]. It has been a hotspot in research area to the impact of tuberculosis and anti-tuberculosis therapy on the composition and modification of human lung microbial communities. However, the relationship between MTBC pulmonary infection and the resident lung microbial composition remains unclear. One study found that the abundance of the Mycobacterium and Porphyromonas genera was increased in TB lesions, while the Cupriavidus genus was reported as dominant and specific in patients with TB [37]. A study showed the difference in bacterial quantity from the URT and LRT, but no difference in composition [41]. Evidence from the other study suggested a significant reduction in alveolar microbiota diversity in BALF samples from patients with TB and interstitial pneumonia, compared to healthy volunteers who took part in the study, which was characterized by a common reduction in Streptococcus and Acinetobacte genus abundances for the two diseases, respectively [42]. It makes sense to investigate the role and diversity of the microbiome and metagenome in mediating MTBC infection. Once the correlation between MTBC infection and microbiota could be established, it will achieve a significant breakthrough in TB research.
The relatively high expense of the mNGS routine (3600 China Yuan, approach 570 dollars) restricts its extensive application and secondary test. Moreover, patients diagnosed with PTB previously were reluctant to receive a second test, which makes it difficult to the supervision and management of drug-resistant MTBC. Few reports were described the alveolar microbiota in BAL samples from patients with TB after anti-TB treatment. More patients will benefit from mNGS when the technique procedure is optimized and the cost is reduced.
This study has several limitations as follows. Firstly, our study involved DNA-based mNGS only; thus, only DNA was sequenced in the study, which means that all RNA viruses were excluded including Influenza virus, Bunyavirus, Hantavirus and so on. Secondly, considering the medical expenses, the enrolled patients did not receive both mNGS analysis of BALF and lung biopsy tissue at the same time. Thirdly, there were some barriers to follow-up patients with PTB because they were required to undergo further treatment in a specialist hospital. Finally, this was a single-center retrospective study. A multicenter, prospective cohort study should be performed in the future.
