Metagenomic next-generation sequencing for the identification of infections caused by Gram-negative pathogens and the prediction of antimicrobial resistance.

Abstract

The aim of this study was to evaluate the efficacy of metagenomic next-generation sequencing (mNGS) for the identification of Gram-negative bacteria (GNB) infections and the prediction of antimicrobial resistance. A retrospective analysis was conducted on 182 patients with diagnosis of GNB infections who underwent mNGS and conventional microbiological tests (CMTs). The detection rate of mNGS was 96.15%, higher than CMTs (45.05%) with a significant difference (χ 2=114.46, P < .01). The pathogen spectrum identified by mNGS was significantly wider than CMTs. Interestingly, the detection rate of mNGS was substantially higher than that of CMTs (70.33% vs 23.08%, P<.01) in patients with but not without antibiotic exposure. There was a significant positive correlation between mapped reads and pro-inflammatory cytokines (interleukin-6 and interleukin-8). However, mNGS failed to predict antimicrobial resistance in 5 of 12 patients compared to phenotype antimicrobial susceptibility testing results. Metagenomic next-generation sequencing has a higher detection rate, a wider pathogen spectrum, and is less affected by prior antibiotic exposure than CMTs in identifying Gram-negative pathogens. The mapped reads may reflect a pro-inflammatory state in GNB-infected patients. Inferring actual resistance phenotypes from metagenomic data remains a great challenge.