SAN FRANCISCO (GenomeWeb) – Interest continues to grow to use metagenomic sequencing as a clinical infectious disease test, given its potential to be fast and agnostic. Last month, researchers from the University of East Anglia's Quadram Institute Bioscience described a nanopore metagenomic sequencing approach for diagnosing lower respiratory illness in a publication on the BioRxiv preprint server. The team is currently testing it in a clinical trial and plans to develop similar methods for diagnosing meningitis and infections caused by joint replacements.
"We're trying to move from proof of concept to clinical implementation of these methods," said Justin O'Grady, a senior lecturer in medical microbiology at the University of East Anglia. "We're in the process of starting to engage with industry to help us develop tests beyond just research-use-only tests."
In the study, the researchers developed a metagenomic sequencing approach that relies on depleting human host DNA and found that sequencing on the Oxford Nanopore Technologies MinIon achieved 96.6 percent concordance with standard culture-based techniques and had a six-hour turnaround time from sample to result compared to several days with standard testing.
O'Grady's team previously published a proof-of-principle study demonstrating the potential for nanopore metagenomic sequencing to diagnose urinary tract infections, but he said that using metagenomic sequencing for UTIs is "kind of like cracking a nut with a hammer" — a bit of an overkill. While in some cases, UTIs can lead to sepsis, the vast majority do not cause serious problems and using metagenomic sequencing to diagnose them would be too expensive, he said.
Instead, the team decided to focus on lower respiratory infections for clinical implementation. These types of infections are often hospital-acquired, are prevalent, and can cause serious problems, he said. In addition, such infections are typically caused only by bacteria and not viruses, which simplifies the testing protocol.
For instance, he said, the protocol makes use of centrifugation, but that would likely lose the virus. "You'd need a slightly different approach," he said. "Even though metagenomics is agnostic once the DNA is in the sequencer, the processing is not agnostic."
Another key to the approach was the researchers' use of a depletion method to get rid of human DNA in order to increase the signal-to-noise ratio of pathogen DNA. For that, they used the chemical saponin, following centrifugation and cell lysis, to break down and wash away the human DNA.
Kristine Wylie, an associate professor at Washington University, who was not involved with the study, but who has developed methods for sequencing viruses in clinical samples, said that the depletion method was key for the metagenomic approach to work. "The big challenge with metagenomics is the fact that pathogens are not super abundant," she said. The researchers demonstrated that the method worked on a "nice range of organisms from clinical samples" of respiratory tract infections.
However, she added that such depletion methods would not be able to be universally applied to all types of infections. For instance, her group has found that for particularly low-abundance organisms, depletion methods don't work because the organism of interest is lost. "Not because the method targets them, but because in the process of sample handling, it just gets lost," she said.
Wylie noted that for the lower respiratory infections tested in this study, the depletion methods were able to get rid of the human DNA without significant loss of bacterial DNA.
In the study, the researchers analyzed a total of 81 clinical samples. Initially, theystudied 40 clinical samples from patients with suspected lower respiratory infection. The first test missed three organisms — Staphylococcus pneumonia, Haemophilus influenza, and Staphylococcus aureus — in three cases. In two of the cases, the patients had a mixed infection and the metagenomic sequencing test missed one of the pathogens.
Next, the team sought to improve the method in order to reduce the false-negative rate. To do this, the researchers focused on the sample prep stage, specifically on cell lysis.
O'Grady said that some Staphylococcus and other organisms are difficult to lyse, so the researchers switched from using a cell lysis buffer to a bead-based approach to physically break open the cells. In a test run of S. aureus, the bead-based approach resulted in 20 times more bacterial DNA.
The second improvement the team wanted to make was to turnaround time. Initially, the protocol took eight hours, including a 1.5-hour human DNA depletion step. The team was able to cut that step down to 50 minutes. It also focused on optimizing the library prep step, reducing the time spent there by one hour.
After optimizing the method, the researchers tested it on another set of 41 clinical samples and achieved a sensitivity of 96.6 percent and a specificity of 41.7 percent.
The metagenomic sequencing approach missed one pathogen that had been detected in routine culture. Again, the sample was diagnosed as a mixed infection by culture, containing Pseudomonas aeruginosa and Escherichia coli, but metagenomic sequencing only detected E. coli. However, the metagenomic technique identified an additional eight pathogens that were missed by culture.
An important aspect of sequencing-based tests is their ability to both identify the causative pathogen and detect antibiotic resistance. In the study, the metagenomic sequencing method identified 184 resistance genes across the 41 samples.
O'Grady said that going forward, one key would be to improve the automated analysis of resistance genes, noting that there were several instances where sequencing detected a resistance gene, but the automated analysis did not call it. Another important step would be to automate the entire protocol to make it an easier-to-use clinical test. Such steps would most likely be done in conjunction with an industry partner, he said. Currently, the process is very hands-on, so it would have to be automated and include appropriate controls. In addition, he said, the bioinformatics pipeline would need refining to get it to a point where it was clinical-grade quality. Then, the goal would be to take it through CE-IVD approval.
"We're looking at both an approach where it is offered as a laboratory-developed test and run as a service, and also developing an actual product," he said.
O'Grady's team has also been involved in a larger clinical trial known as Inhale to study two PCR-based tests running on the Curetis Unyvero and the BioMérieux BioFire Diagnostics FilmArray along with metagenomic sequencing on the MinIon for diagnosing respiratory illness.
The trial, which is funded by the UK's National Institute for Health Research, is a collaboration between the University of East Anglia, University College London, and four hospitals — Norwich & Norfolk University Hospitals, University College London Hospitals, Great Ormond Street Hospital Children's Charity, and BUPA Cromwell Hospital.
The group is nearing the completion of the first part of the trial, he said, which has included performing metagenomic sequencing on around 300 clinical samples of hospital-acquired pneumonia using the method described in the BioRxiv paper.
The goal is to compare the three tests to each other and to culture-based diagnostics with regard to performance, patient outcome, the impact on antibiotic prescriptions, and costs.
O'Grady declined to disclose any preliminary results from the trial but said he anticipated that they would be published either toward the end of this year or early next year.