Real-time metagenomics for rapid clinical pathogen identification and resistance prediction in gram-negative bloodstream infections

Abstract

Bloodstream infections have become one of the main causes of hospital mortality with an estimated five million deaths per year globally. Recent trends have shown an increase in the proportion of multi-drug resistant gram-negative bacteria as causative agents. These strains represent the biggest challenge in treating bloodstream infections, as resistant infections are associated with higher mortality and longer hospital stays due to delayed and/or inappropriate therapy. The inappropriate and often broad-spectrum empirical therapy, in turn, increases selection pressure for resistant pathogens perpetuating the rise in antibiotic resistance.

A rapid and accurate identification of the infection causing pathogen and its resistance patterns would facilitate early administration of appropriate antibiotic therapy leading to reduced resistance selection pressure and decreased mortality and morbidity. However, bloodstream infection diagnosis is currently mainly culture-based, which is typically time- and resource-consuming and has low clinical sensitivity and high culture bias. In this study, we have therefore started implementing metagenomic approaches using nanopore sequencing to overcome the limitations of culture-based approaches by making accurate real-time resistance predictions of all relevant bacterial strains to ultimately improve patient outcome. This study is conducted at the university hospital of TUM Rechts der Isar in Munich, Germany, since hospital mortality of sepsis patients in Germany is notably higher than in other high-income countries. We will here present our first results of sequencing DNA from routine blood cultures of ICU patients at TUM Rechts der Isar, and compare the accuracy and the turnaround time of our real-time genomics approach to the current state-of-the-art culture-based methods. Our preliminary results show a promising overlap in taxonomic and functional predictions with culture-based approaches, for example with respect to Klebsiella pneumoniae infection prediction and assessments of intrinsic Ampicillin and Piperacillin resistance.

Our long-term aim is to develop fully integrated genomic and computational solutions to bring about an accurate, data-driven, and cost-effective approach for fast in-depth assessments of bloodstream infections directly at the point of care. This project has the potential to maximise the efficiency of antimicrobial therapy reducing the use of broad-spectrum antimicrobials and increasing survival rate in the context of bloodstream infections.