Early stage active TB disease is often missed by diagnosticians owing the absence of symptoms of TB. To address this issue, radiological and immunological testing of sputum or BALF may be conducted for detecting TB cases. Nevertheless, clinicians are usually unwilling to initiate early empiric anti-TB therapy simply on the basis of radiological and immunological test results when patients are unable to produce sputum. Thus, there is an urgently needed to develop new diagnostic techniques to help clinicians diagnose sputum-scarce TB patients correctly.

As compared with previous studies demonstrating good pathogen detection results of clinical sputum samples based on nanopore assay, here we assessed the diagnostic value of BALF samples using nanopore sequencing, a method that to date has mainly been applied to genomic DNA sequencing [11]. With advances in sequencing chemistry and computational power, in recent years, nanopore sequencing assay has been increasingly used for clinical applications [12]. For example, such assays have been extensively used to diagnose patients with fever who lack TB culture-positive results and detectable localised infections [13, 14] and have enabled rapid diagnosing of patients with slow-growing microorganisms-caused infections by such as TB and non-tuberculous mycobacteria (NTM) [15,16,17,18,19,20].

It is of great importance to rapidly identify mycobacteria at the species level to differentiate between TB and NTM infections in order to select appropriate medications for patient treatment that can minimise unnecessary testing and unnecessary drugs-related side effects, reduce overall cost of therapy and maximise treatment outcomes. For this reason, sequence-based assays requiring at most a few days to complete, and it greatly affects the clinical diagnosis of TB, as clinical management of mycobacterial infections that has traditionally depended on sample testing by private laboratories [2]. In fact, nowadays a single sequencing run can process more than 10 samples, dramatically reducing the cost of sequencing individual samples, which was prohibitively high at the time nanopore sequencing assays were initially utilized to diagnose infectious diseases [2]. Indeed, the MinION sequencing device developed by ONT has become increasingly popular due to its advantages, such as low device cost, short run time and small size that easy portability [7, 8] that have enabled increasing numbers of clinical laboratories to adopt this platform for use in conducting diagnostic nanopore sequencing assays. In 2015, the device was first made commercialized, it had a high sequencing error [21]. However, after the device underwent several rounds of re-engineering, its sequencing error rate decreased to an acceptable level thus the market demand for the device surged [22, 23]. In turn, this increase in demand has boosted efforts to further improve device scalability, convenience and flexibility that have enabled the improved MinION devices to serve clinical microbiology laboratories better and dramatically transform clinical microbiological testing [2].

The detection of fungi, bacteria, parasitic organisms and viruses has been facilitated due to the next-generation sequencing through untargeted DNA/RNA sequencing, which has enabled rapid pathogen identification to support the accurate diagnosis of infectious diseases at their earliest stages. In patients with pulmonary infections whose pathogens were not detected using traditional pathogen detection methods, a sequence-based assay developed by Huang et al. using this technology, that allowed in 94.49% samples successful detection of human pathogens. Moreover, their results showed that sequence-based tests were more accurate and sensitive than assays using standard pathogen detection [24]. In the mixed pulmonary infections detection, Wang et al. also identified sequence-based detection methods to be more sensitive than traditional methods [25]. In present study, we found that nanopore sequencing of PCR products enabled us to effectively diagnose suspected PTB cases, while also providing superior sensitivity (83.33%) as compared to sensitivities obtained with MGIT culture (36.11%) and Xpert MTB/RIF (40.28%). Furthermore, in the diagnosing of suspected PTB cases, in comparison with MGIT culture and Xpert MTB/RIF analysis, the nanopore sequencing had better performance in the diagnosis, as reflected by Youden’s index and area under the curve (AUC) values. In summarizing, above results indicate that nanopore sequencing holds promise to be a valuable additional assay to optimize the diagnostic detection of PTB cases, especially for the detection of TB and NTM organisms that cannot be distinguished the MGIT culture assay and thus produced positive MTB culture results for both mycobacterial types, with 3 of 16 NTM-containing control samples testing positive for MTB. For the 3 cases, PNB and CT were used to identify the possibility of co-infection and the effectiveness of treatment options. The PNB identification method is an effective microbiological tool that utilizes the fundamental differences between TB and NTM in metabolizing PNB. Through this method, these two types of bacteria can be effectively distinguished. On PNB selective medium, the growth of TB is inhibited, while NTM is able to grow, thus achieving separation and identification of the two. This high-number of false positive outcomes corroborated the low specificity of the MGIT culture test and indicated that when using this test for TB diagnosis in NTM epidemic areas, the results it provides should be interpreted thoroughly. Nevertheless, the specificity of MGIT culture assay was adapted to differentiate between mycobacterial and non-mycobacterial organisms.

With regard to assay specificity, Xpert only specifically targets the MTB complex-associated rpoB sequence, which exists in only one copy within the MTB genome. In contrast, through targeting the IS6110 insertion sequence (which is not present in NTM genomes) and thus nanopore sequencing analysis has specific for the M. tuberculosis complex. In addition, the IS6110 sequence is present at 10 to 12 copies per genome for diverse MTB strains and thus nanopore sequencing based on detection of this sequence can provide superior MTB-detection sensitivity when used to detect diverse MTB isolates that exhibit different tissue spread patterns and pathogenicity. In order to confirm the reliability of the nanopore sequencing results, we also used hsp65 and gyrB genes as the basis for distinguishing the Mycobacterium tuberculosis complex and NTM.

In addition, the limitations of this study include: firstly, due to the smaller sample size for this study, so the results obtained may be biased and therefore future diagnostic potency of nanopore sequencing studies should be investigated using a larger sample size. Second, the focus of this study was the detection of BALF samples, so lavage samples collected containing pathogen numbers below the assay lower limit of detection could lead to false negative results. Therefore, nanopore sequencing should still be viewed as an auxiliary diagnostic tool to be apply to conjunction clinical characteristics, radiology imaging results and other laboratory testing findings. Third, false positive and false negative nanopore sequencing analysis results cannot be excluded which caused by (1) the depths of sequence were too low; (2) the biomass of microbial pathogen was low while the background noise of host genome was high; (3) the patient was on antibiotics before the test; and (4) the samples were contaminated with human flora or environmental microorganisms [25]. We checked the specificity of the primer and found that it performed well in the identification of mycobacteria. The samples were loaded with human flora and environmental microorganisms. This was mainly because the background microorganisms in the samples were too rich, which made the amount of Mycobacterium tuberculosis nucleic acid (when the PCR input was 2–20 ng) relatively low, which may lead to false negative results. We will also optimize our experimental process in the future.Fourth, although one patient’s PCR products contained an antimicrobial drug resistance gene sequence, we did not obtain the patient’s antibiotic treatment result and thus could not determine whether the sequencing results of patients are aligned with treatment response. Fifth, patients with certain control diseases, like rheumatoid arthritis and lymphomas, were absence in our study cohort, which may have biased the results. However, the shell vial culture and nanopore sequencing assays of biopsied tissue samples could overcome this issue. Sixth, this study found that one patient’s PCR products contained an antimicrobial drug resistance gene sequence, and the role of nanopore sequencing in the direct detection of rifampicin and fluoroquinolone resistance was not evaluated. In future, a few drug-resistant smear negative BAL samples should be included to further investigate the potential application of nanopore sequencing in drug susceptibility testing in further studies. Compared with MGIT culture and Xpert MTB/RIF assays, BALF’s nanopore sequencing provided superior MTB detection sensitivity and thus is suitable for testing of sputum-scarce suspected PTB cases, which be used as a complement to MTB testing. Nevertheless, our findings need to be validated by including a larger and more diverse patient population for further studies.