NEW YORK(GenomeWeb) —University of Utah and ARUP Laboratories researchers have shown in a small study that direct metagenomic sequencing of clinical samples correlates well with a standard panel assay used in clinical respiratory virus testing.
The results, published in a poster at last month's annual meeting of the Association for Molecular Pathology, suggest that a metagenomics approach to respiratory virus detection has high sensitivity that may even surpass standard methods of detecting co-infections of more than one pathogen.
Erin Graf, a postdoc at ARUP and first author of the poster, told GenomeWeb that she and her colleagues at first expected metagenomics to be less sensitive than something like a PCR or other amplification-based technique.
"But it turned out that in our hands at least, it was every bit as sensitive and allowed us to detect more things beyond what the respiratory virus panel we use can look for," she said.
In their study, Graf and coauthors directly compared a metagenomics approach using Illumina's HiSeq 2500 to the standard respiratory virus test used in their lab, GenMark's eSensor Respiratory Virus Panel.
The team collected a total of 67 nasopharyngeal swab samples from children under 5 submitted for RV testing between April 2013 and March 2014 and supplemented this cohort with a second retrospective set of 35 samples that were known-positive by the GenMark assay and enriched for the presence of dual infections.
The group found that the GenMark RSV assay and the metagenomic sequencing approach were largely concordant overall in the prospectively collected set of 67 swab samples. However, metagenomics identified more than 10 viruses that the panel did not. The GenMark panel also detected three viruses that the metagenomic method did not — one adenovirus, one rhinovirus, and one case of RSV-B.
Graf said that when the team used a third, lab-developed PCR strategy to reassess the discordant samples, it confirmed the three negative metagenomic calls and not the GenMark positives. Whether that means that the panel calls were false positives or cases of very low-level viral infection can't be known, she said.
Looking at the additional 35 known-positive samples, the group also found that metagenomics detected more co-infections than the GenMark panel, Graf said.
The clinical implications of this are less clear at this point, she said. Detecting co-infection may be useful in clinical practice, as some data has shown that dual infections are more severe, but this is not yet definitive.
The team also reported that their results indicated that contamination is a significant concern for metagenomic viral detection, though they were able to map contaminating reads back to each sample to distinguish true infections in the study.
According to Graf, though clinical use of metagenomics on human samples is still in its infancy, wider use doesn't look so far off, as the barriers are rapidly falling.
She cited a study earlier this year in the New England Journal of Medicine in which researchers from the University of California, San Francisco successfully diagnosed a difficult bacterial infection using metagenomics.
In that paper, the UCSF group reported a case study in which unbiased next-gen sequencing of the cerebrospinal fluid of a 14-year old boy identified a leptospira infection even as clinical assays for the pathogen were negative.
Subsequent treatment for leptospirosis was successful. And later PCR and serologic testing at the US Centers for Disease Control and Prevention confirmed the metagenomics result.
In addition, researchers from theUniversityofWarwickin theUKalso published a study in September demonstrating that direct metagenomic sequencing of sputum samples could accurately detect tuberculosis infection.
The approach is also garnering commercial interest: Last month, a new company called Aperiomics launched metagenomic sequencing and analysis services.
As for the ARUP work, Graf said her departure from the laboratories will prevent her from continuing her own study with the same team of coauthors, but said that there are a number of important avenues that the field may be interested in following up.
For example, "one piece we didn't start looking at in my group is bacterial contribution in these samples," she said. "That analysis is more complicated because the respiratory system is colonized by bacteria, so picking apart colonizations from pathogens can be a challenge. Also, the genomes are much larger than virus sequences and have a lot of homology, so speciating bacteria in these samples takes a lot more work."
Fungi also present another promising target for metagenomics, she added.
From there, a larger goal would be to push even farther the bounds of discovery in a single human sample. Beyond detecting pathogens, unbiased metagenomics also has the potential to yield information from human host nucleic acids.
"Another obvious step labs are interested in is looking at post-transcription profiling to measure how the human RNA changes in the face of particular pathogens. That's also something you could do with the one sample that we have," Graf said.