NEW YORK (GenomeWeb) – A team led by investigators at the Chan Zuckerberg Biohub and the University of California, San Francisco has developed a sequencing-based strategy for simultaneously assessing pathogen, microbiome, and host response features to potentially diagnose individuals with lower respiratory tract infections.
"This study suggests that a single streamlined protocol offering an integrated genomic portrait of pathogen, microbiome, and host transcriptome may hold promise as a tool for LRTI diagnosis," co-senior authors Joseph DeRisi, a researcher affiliated with CZ Biohub and UCSF, and Carolyn Calfee, a pulmonary, critical care, allergy, and sleep medicine researcher at UCSF, and their colleagues wrote.
Despite the prevalence, health impacts, and mortality risk associated with lower respiratory tract infections around the world, the team explained, it often remains difficult to identify the specific pathogens behind these infections due to the presence of non-pathogenic microbes present in the inflamed lung environment.
"Early and accurate determination of acute respiratory disease etiology is crucial for implementing effective pathogen-targeted therapies," the authors wrote, "but is often not possible due to the limitations of current microbiologic tests in terms of sensitivity, speed, and spectrum of available assay targets."
As they reported online yesterday in the Proceedings of the National Academy of Sciences, DeRisi, Calfee, and colleagues focused on 92 individuals with acute respiratory failure who were enrolled prospectively at the UCSF Moffitt-Long Hospital intensive care unit from summer 2013 to fall 2017. The group included 26 individuals with lower respiratory tract infections and 18 without.
In an effort to identify informative pathogen sequences, microbiome features, and host gene expression patterns that characterized the lower respiratory tract infection patients, the team did metagenomic DNA and RNA sequencing on tracheal aspirate samples from the patients. From there, the group turned to rules-based and logistic regression computational models aimed at assaying for lung pathogens based on the DNA and RNA features gleaned from the metagenomic data.
The researchers initially tested the algorithms on 20 patients with or without acute lower respiratory tract infections, who had relatively clear diagnoses from the outset. They subsequently validated the models using data for another two dozen individuals, where they could detect lower respiratory tract infections with nearly 96 percent accuracy before taking the approach forward to a larger group of patients.
"[W]e address the need for better [lower respiratory tract infection] diagnostics by developing an [metagenomic next-generation sequencing]-based method that integrates host response and unbiased microbe detection. We then evaluate the performance of this approach in a prospective cohort of critically ill patients with acute respiratory failure," the authors wrote.
From these data, they suggested that it is possible to bring together pathogen, microbiome diversity, and host gene expression metrics to identify LRTI-positive patients and differentiate them from critically ill controls with non-infectious acute respiratory illnesses.