Heliyon (Jan 2024)
Metagenomic next generation sequencing of bronchoalveolar lavage samples for the diagnosis of lower respiratory tract infections: A systematic review and meta-analysis
Abstract
Background: Lower respiratory tract infections (LRTI)are known to be diagnosed late or inaccurately. This has fueled the unscrupulous use of antibiotics, as they are often used empirically and clinically, leading to antibiotic abuse and multidrug resistance in patients. Metagenomic next-generation sequencing (mNGS), now widely used in clinical studies, could be a potential intervention to revolutionize microbiology by rapidly identifying unknown species. Methods: This review and meta-analysis were conducted on eligible studies with respect to metagenomic sequencing on clinical LRTI diagnostics up to May 01, 2022. QUADAS-2 was employed to assess the methodological bias as well as applicability. After that, a meta-analysis was conducted to analyze the accuracy of mNGS, compared with the composite reference standard (CRS), among the enrolled studies. Results: This work collected 1248 samples in 13/21 qualified articles to factor in the accuracy of the diagnostic test. Typically, methods like molecular testing, culture, composite measures, and clinical decision-making were adopted as the reference criteria. With regard to Bronchoalveolar Lavage Samples, their sensitivity was 89% (82–93%) while their specificity was 90% (66–98%), with obvious heterogeneities in these two factors as demonstrated by different studies. The summary receiver operating characteristic (SROC) curve was plotted for mNGS as a function of LRTI, and the area under the curve (AUC) was 0.94. A Funnel plot with a p-value greater than 0.05 indicated the absence of publication bias. Positive and negative likelihood ratios (PLR and NLR) were >10 and > 0.1, respectively. In this pre-test probability-post-probability-likelihood ratio relationship graph, the values were Prior prob (%) = 20, Post-prob-Pos (%) = 77 and Post-prob-Neg (%) = 4. Conclusion: The AUC value of SROC suggested a high accuracy of mNGS in diagnosis, with no publication bias and high reliability. The application of mNGS exhibits notable diagnostic efficacy in discerning pathogens present in bronchoalveolar lavage fluid (BALF) among patients afflicted with LRTI. However, mNGS is more meaningful for the definitive diagnosis of the disease rather than the exclusion of the disease. This post-test probability is significantly higher than the pre-test probability.
