Diagnostic accuracy of the metagenomic next-generation sequencing (mNGS) for detection of bacterial meningoencephalitis: a systematic review and meta-analysis

Abstract

The early diagnosis of bacterial meningoencephalitis (BM/E) is difficult, and delay in diagnosis can cause complications leading to neurological impairment/death. In cases of unexplained BM/E, the metagenomic NGS (mNGS) offers an advantage over conventional methods, especially when a rare pathogen is implicated or the patient is on antibiotics. This study aims to evaluate and compare the diagnostic efficacy of mNGS for the diagnosis of BM/E using cerebrospinal fluid (CSF) specimens versus a composite reference standard (CRS). The electronic databases (Embase, PubMed, and Web of Science) were searched up to 15 June 2021. Studies such as cohort, case–control, prospective, or retrospective studies that assessed the diagnostic efficacy of mNGS in suspected bacterial meningitis/encephalitis cases were included. Ten studies met the inclusion criteria, including three retrospective and seven prospective studies. The sensitivity of mNGS for diagnosis of BM/E from CSF samples ranged from 33 (95% CI: 13–62) to 98% (95% CI: 76–99). The specificity of mNGS ranged from 67 (95% CI: 55–78) to 98% (95% CI: 95–99). The estimated AUC (area under curve) by hierarchical summary receiver operating characteristic (HSROC) of the studies being analyzed was 0.912. The meta-regression analysis demonstrated that the different types of studies (single-center vs. multi-center) had an effect on the specificity of mNGS for BM/E compared with CRS (90% vs. 96%, meta-regression P < 0.05). The current analysis revealed moderate diagnostic accuracy of mNGS. This approach can be helpful, especially in cases of undiagnosed BM/E by identification of organism and subsequently accelerating the patient management.