TABLE 2.
Effectiveness of metagenomic next-generation sequencing (mNGS) for MTB detection.
| References | Research type and sample size | Research conclusions and results |
| Li et al. (46) | • Type: A * • Sample: lung tissue (n = 20) • Pre-ATT samples (not mentioned) • MTBC mNGS positive (n = 4) • Pathogenic TB diagnoses (n = 4) Diagnosis basis: AFS, Xpert, PCR, etc. |
• MNGS showed the highest Spe and PPV for MTBC when compared with histopathology method. • For MTBC lung tissue mNGS: compared with smear: Sen: 100.0% (19.8–100.0%), Spe: 88.9% (63.9–98.1%); compared with histopathology: Sen: 100.0% (31.0–100.0%), Spe: 94.1% (69.2–99.7%). |
| Miao et al. (6) | • Type: C • Sample: all samples (n = 511) • Pre-ATT samples (not mentioned) • TB mNGS positive (n = 42) • Total TB diagnoses (n = 92) Pathogenic TB diagnoses (n = 25) Diagnosis basis: culture, clinical criteria |
• MNGS outperformed culture, especially for MTB [odds ratio = 4 (1.7–10.8)]. • For MTB mNGS: Sen: 45.7% (42/92). • For NTM mNGS: Sen: 29.8% (17/57). |
| Wang et al. (51) | • Type: A + B • Sample: CSF (n = 29)$ • Pre-ATT samples (not mentioned) • TB mNGS positive (n = 42) • Total TB diagnoses (n = 23) Pathogenic TB diagnoses (n = 12) Clinical TB diagnoses (n = 11) Diagnosis basis: culture, AFS, PCR, clinical criteria |
• Combining mNGS and conventional methods (culture, AFS, PCR) increased the detection rate to 95.65%. • For MTBC CSF mNGS: Sen: 66.67% (8/12); Spe: 100% (6/6); PPV: 100% (8/8); NPV: 60% (6/10); accuracy: 77.78% (14/18). |
| Zhou et al. (62) | • Type: B • Sample: all samples (n = 105) Pulmonary samples (n = 27) CSF samples (n = 49) Other extrapulmonary samples (n = 29) • Pre-ATT samples (n = 27) • TB mNGS positive (n = 20) • Total TB diagnoses (n = 45) Diagnosis basis: culture, Xpert, PCR, clinical criteria |
• Combining mNGS and Xpert improved the etiology diagnosis, increased specificity from 44% (20/45, 30–60%) to 60% (27/45, 44–74%); • Empirical ATT reduced diagnostic efficacy of culture, Xpert, and mNGS. • For MTB mNGS: Sen: 44% (20/45, 30–60%), Spe: 98% (59/60, 91–100%); • For MTB pulmonary samples mNGS: Sen: 62% (8/13, 32–86%); Spe: 100% (14/14, 77–100%); • For MTB CSF mNGS: Sen: 44% (7/16, 20–70%); Spe: 97% (32/33, 84-99%); • For MTB other extrapulmonary samples mNGS: Sen: 31% (5/16, 11–59%); Spe: 100% (13/13, 75–100%). |
| Xing et al. (77) | • Type: A • Sample: CSF (n = 213) Pre-ATT samples (not mentioned) • TB mNGS positive (shown rightward) • Total TB diagnoses (n = 44) Pathogenic TB diagnoses (n = 6) Clinical TB diagnoses (n = 38) Diagnosis basis: AFS, Xpert, clinical criteria |
• When the genus-specific read number ≥ 1 was considered MTB positive, the AUC (61.9%, 51.6–72.1%) was largest. • Given high specificity (96.4%, 163/169) of mNGS in the diagnosis of TBM, it allows a negative mNGS test to be used as one of the diagnostic methods to exclude TBM. • For MTB CSF mNGS: if genus-specific read numbers ≥ 1, 2, 3, 5, 10 was considered positive; the positive consistency rates were 27.3, 20.5, 18.2, 13.6, 6.8%; the negative consistency rates were 96.4, 97.6, 98.2, 99.4, and 100%; the total consistency rates were 82.2, 81.7, 81.7, 81.2, and 80.8%, respectively. |
| Yan et al. (50) | • Type: A + B • Sample: CSF (n = 51) • Pre-ATT sample (n = 51) • TB mNGS positive (n = 38) • Total TB diagnoses (n = 45) Pathogenic TB diagnoses (n = 38) Clinical TB diagnoses (n = 7) Diagnosis basis: culture, AFS, PCR, Xpert, clinical criteria (102) |
• Patients with a significant increase in CSF cell number and protein quantification might have a higher likelihood of positive MTB detection of mNGS. • For MTB CSF mNGS: Sen: 84.44% (38/45, 69.94–93.01%); Spe: 100% (6/6, 51.68–100%); PPV: 100.0% (40/40, 88.57–100%); NPV: 46.15% (6/13, 20.40–73.88%). |
| Chen et al. (70) | • Type: B. • Sample: all samples (n = 70) Pulmonary samples (n = 37) Extrapulmonary samples (n = 33) • Pre-ATT samples (not mentioned) • TB mNGS positive (n = 25) • Total TB diagnoses (n = 36) Pathogenic TB diagnoses (n = 36) Diagnosis basis: pathological test, PCR |
• Combining mNGS and culture or Xpert improved Sen to72.2% (26/36, 54.6–85.2%), higher than only mNGS (66.7%, 24/36, 48.9–80.9%), showing the potential for clinical application in TB. • For MTB pulmonary samples mNGS: Sen: 82.4% (14/17, 55.8–95.3%); Spe: 100% (20/20, 80.0–100.0%); PPV: 100% (14/14, 73.2–100.0%); NPV: 87.0% (20/23, 65.3–96.6%); Youden index: 82.4%; • For MTB extrapulmonary samples mNGS: Sen: 47.4% (9/19, 25.2–70.5%); Spe: 92.9% (13/14, 64.2–99.6%); PPV: 90.0% (9/10, 54.1–99.5%); NPV: 56.5% (13/23, 34.9–76.1%); Youden index: 40.3%. |
| Jin et al. (30) | • Type: B. • Sample: all samples (n = 820) Pulmonary samples (n = 477) Extrapulmonary samples (n = 343) Pre-ATT samples (not mentioned) • TB mNGS positive (n = 76) • Total TB diagnoses (n = 125) Pathogenic TB diagnoses (n = 64) Clinical TB diagnoses (n = 61) Diagnosis basis: culture, Xpert, PCR, clinical criteria |
• mNGS may be a promising technology for sputum-negative PTB and tuberculous serous effusion. • For MTB mNGS: Sen: 49.6% (62/125, 40.6–58.6%), Spe: 98.3% (683/695, 96.9–99.1%); • For MTB pulmonary samples mNGS: Sen: 58.5% (31/53, 44.2–71.6%); Spe: 98.3% (417/424, 96.5–99.3%); • For MTB extrapulmonary samples mNGS: Sen: 43.1% (31/72, 31.6–55.2%), Spe: 98.2% (266/271, 95.5–99.3%). |
| Shi et al. (61) | • Type: B • Sample: BALF (n = 110) • Pre-ATT samples (not mentioned) • TB mNGS positive (n = 24) • Total TB diagnoses (n = 48) Pathogenic TB diagnoses (n = 32) Clinical TB diagnoses (n = 16) Diagnosis basis: culture, AFS, Xpert, clinical criteria |
• mNGS identified 67.23% infection cases within 3 days, while the conventional methods identified 49.58% infection cases for over 90 days. • For MTB BALF mNGS: Sen: 47.92% (23/48, 33.5–62.6%), similar to that of Xpert (45.83%, 22/48) and culture (46.81%, 22/47), but much higher than that of AFS (29.17%, 14/48); Spe: 98.39% (61/62, 90.2–99.9%). |
| Sun et al. (71) | • Type: B • Sample: smear-negative extrapulmonary samples (n = 208) • Pre-ATT samples (n = 129) • TB mNGS positive (n = 101) • Total TB diagnoses (n = 180) Clinical TB diagnoses (n = 180) Diagnosis basis: clinical criteria |
• mNGS is superior for TB on smear-negative extrapulmonary specimens and could identify all possible pathogens within 48 h; mNGS positive rate was highest for TBM (84.44%, 38/45). • For MTB smear-negative extrapulmonary samples mNGS: Sen: 56.11% (101/180, 48.53–63.43%), Spe: 100% (28/28, 84.98–100.00%); PPV: 100% (101/101, 95.43–100.00%); NPV: 26.17% (28/107, 18.36–35.71%). |
| Liu et al. (72) | • Type: A • Sample: BALF (n = 322) • Pre-ATT TB samples (n = 142) Post-ATT TB samples (n = 69) • MTBC mNGS positive (n = 118) • Total TB diagnosis (n = 211) Diagnosis basis: culture, AFS, Xpert, clinical criteria |
• Positive MTBC detection by mNGS was affected by Vitamin D, TB initial treatment/retreatment, erythrocyte sedimentation rate and cavity in chest imaging, but not by prior ATT within 3 months. • For MTBC pre-ATT BALF mNGS: Sen: 59.9% (85/142); while for MTBC post-ATT BALF mNGS: Sen 47.8% (33/69). |
| Lin et al. (103) | • Type: A + B • Sample: CSF (n = 50) • Pre-ATT samples (not mentioned) • MTBC mNGS positive (n = 20) • Total TB diagnosis (n = 34) Pathogenic TB diagnoses (n = 22) Clinical TB diagnoses (n = 12) Diagnosis basis: culture, AFS, Xpert, clinical criteria |
• mNGS could rapidly detect MTBC in CSF, which could be used as an early diagnosis index of TBM. mNGS combined with MTB culture could increase the detection rate. • For MTB CSF mNGS: Sen: 58.8% (20/34), Spe: 100.0% (16/16). |
| Zhu et al. (104) | • Type: B • Sample: BALF (n = 78) • Lung tissue (n = 29) • Pre-ATT samples (not mentioned) • TB mNGS positive (n = 43) • Total TB diagnosis (n = 46) |
• mNGS offers improved detection of MTB in BALF or lung tissue biopsy samples in sputum-scarce or smear-negative cases. • For MTB BALF mNGS: Sen: 90.63% (29/32, 73.83–97.55%), Spe: 97.83% (45/46, 87.03–99.89%), PPV: 96.67% (29/30, 80.95–99.83%); NPV: 93.75% (45/48, 81.80–98.37%); • For MTB lung tissue mNGS: Sen: 85.71% (12/14, 56.15–97.49%); Spe: 93.33% (14/15, 66.03–99.65%), PPV: 92.31% (12/13, 62.09–99.60%), NPV: 87.50% (14/16, 60.41–97.80%). |
*Given the different focuses in different researches, the inclusive criteria varied. We classified the published mNGS literature into three types according to research focus and inclusive criteria: Type A (specific sample type was included, such as lung tissue, BALF, or CSF), Type B (specific pathogen was included, such as MTB), and Type C (comprehensive studies that enrolled all samples or patients in the research organizations). This review only includes the parts related to MTB infection.
$It clearly defined the control groups: positive controls (bacterial/cryptococcal meningitis or viral meningoencephalitis) and negative control (auto-immune encephalitis).
Although there was no specific definition in other studies, similar study designs were carried. Therefore, control was not mentioned in this review.