By Deborah Borfitz
January 6, 2021 | An all-in-one metagenomics test developed by University of California San Francisco (UCSF) scientists could transform the way infectious diseases are diagnosed at the point of care, irrespective of the sample type, according to Charles Chiu, M.D., Ph.D., a professor in the UCSF department of laboratory medicine and director of the UCSF-Abbott Viral Diagnostics and Discovery Center. The same technology is now being adapted as a means to simultaneously detect SARS-CoV-2 and track down drivers of COVID-19 outbreaks at the community level.
Metagenomic next-generation sequencing has been around for more than a decade and the idea behind it is “really simple,” says Chiu. “It’s essentially a shotgun sequencing approach of all the DNA and RNA in a sample… a needle-in-the-haystack endeavor where we’re trying to identify the very small fraction of nonhuman sequences within clinical samples.”
The challenging part has traditionally been bioinformatics interpretation of test results, Chiu says, because samples collected from the human body have their own microbiome inclusive of bacteria and viruses that could be colonizers or contaminants rather than true pathogens. But the new clinical laboratory test can zero in on the microbial miscreant afflicting hospitalized patients with serious infections in as little as six hours.
It matters not the type or species of infectious agent or if doctors have an inkling as to the culprit, he continues. The new test relies on specially developed analytical software to compare DNA sequences in the sample to massive genomic databases covering all known pathogens to identify a match. The technique works for all body fluids—blood, urine, stool, spinal, abscess, pleural, peritoneal, joint, tonsillar, and vitreous (eyeball)—without any special handling or processing required.
Current clinical practice is to test many different sample types when attempting to make a diagnose because it is often unclear which type is best, says Chiu. “The idea with metagenomic testing is that instead of sending off many specific tests for each sample type, why not just send off one test that we can use universally on all body fluids?”
The metagenomic testing platform uses cell-free DNA in samples and a dual-use barcoding protocol to identify pathogens and, in a study recently published in Nature Medicine (DOI: 10.1038/s41591-020-1105-z), performed at least as well as bacterial and fungal polymerase chain reaction (PCR) assays. The comparison was made on 182 body fluids from 160 patients with acute illness.
Notably, the test proved to be compatible with both a traditional sequencer (Illumina, U.S.) and a pocket-sized sequencer (Oxford Nanopore Technologies, U.K.) powered by the USB port of a laptop computer. “That expands the potential applications of the technology to low-resource settings in the developing world and smaller hospital labs that don’t have the sophisticated equipment and computational power needed [for traditional DNA sequencing],” says Chiu, as well as point-of-care sites such as emergency rooms, outpatient clinics, homes, and hotel rooms.
Using Illumina sequencing, test sensitivity and specificity of detection were 79% and 91% for bacteria and 91% and 89% for fungi. Using nanopore sequencing, the figures were, respectively, 75% and 81% for bacteria and 91% and 100% for fungi.
Most of the six hours it currently takes to get from sample to answer is consumed by generating the sequencing libraries on giant biorobots, continues Chiu. The research team currently has several microfluidics companies working on a portable cartridge that would provide results at the push of a button, as the goal is to develop this into a point-of-care test approved for a CLIA waiver, he adds. “It’s more of an engineering problem now, not a problem with analyzing the data or even the interpretation.”
A clinical microbial sequencing board, modeled after tumor boards in oncology, can be convened to discuss the significance of test results with ordering physicians, says Chiu. The engineering challenge is in automating the library prep step, which he expects can be readily trimmed to two or three hours but remains the “huge limiting factor.” Isothermal amplification methods for generating the library are now being explored in hopes of reducing that step to an hour or less.
To function as a real-world clinical laboratory test will also require automating results reporting with an attractive graphical visualization interface, he adds, and ideally also on-the-spot interpretation. “We’re not quite there yet, but the pipeline itself that analyzes the sequence data is fairly automated... from sample to answer.” Data analysis happens wholly by computer in the cloud.
Unmet Clinical Needs
The experimental all-in-one metagenomics test meets the typical benchmark of having a sensitivity within 80% to 90% of culture, the diagnostic gold standard for many bacterial and fungal infections, and is “arguably better,” says Chiu. For the Nature Medicine study, the research team worked with a lot of weak culture-positive cases grown from an enrichment broth, which suggests the original levels of pathogen in the samples were very low.
Another possibility, he adds, is that some of the samples were actually contaminants and “we really can’t tell the difference.” In other words, the new test was being compared against an imperfect gold standard. At least part of the reason for not getting to 100% sensitivity is due to false-positive culture results, Chiu says.
Of particular note is that the rapid metagenomics test diagnosed infections in seven of 12 patients whose illnesses couldn’t be identified by the culture and PCR methodology, Chiu says. This suggests it can address an unmet clinical need, although a prospective study will be needed to confirm that the method has better diagnostic yield than currently available options.
Metagenomic sequencing is already being done on specific sample types, says Chiu, pointing to a spinal fluid test for diagnosing neurologic infections in patients with meningitis and encephalitis that he developed at UCSF and was used in a study published last year in the New England Journal of Medicine (DOI: 10.1056/NEJMoa1803396). Similarly, Karius (Redwood City, California) has a metagenomic test for plasma and IDbyDNA (Salt Lake City, Utah) has one for respiratory fluids.
One of the current limitations of the all-in-one metagenomics is that, unlike the spinal fluid test, it doesn’t look at RNA as well as DNA, he continues. That means it can’t detect RNA viruses like SARS-CoV-2, Ebola, Zika, H1N1, MERS, and influenza.
Although the research team has already developed an RNA version of the test, generating two sequencing libraries increases turnaround time, explains Chiu. That said, nearly all of the novel viruses implicated in outbreaks over the past decade come from known viral families, making them identifiable by metagenomic sequencing.
In 2012, Chiu and his colleagues identified the novel Bas-Congo virus (PLOS Pathogens, DOI: 10.1371/journal.ppat.1002924) based on the fact that it shares 30% of the nucleotides of known viruses in the Rhabdoviridae family that includes the virus that causes rabies. Similarly, SARS-CoV-2 can be identified in a sample based on sequencing information that is “exquisitely specific” to the coronavirus family. “That suggests to me that as long as any novel pathogen or virus has some sufficient similarity [to a known close relative]… we can identify it,” he says.
EUA Pathway
Metagenomic sequencing by primer enrichment, an adaptation of the same technology (Nature Microbiology, DOI: 10.1038/s41564-019-0637-9), was recently used to show that at least seven strains of SARS-CoV-2 entered Northern California in 2020, as described a few months ago in Science (DOI: 10.1126/science.abb9263). Enriching the library for the SARS-CoV-2 sequences allowed the researchers to not only detect COVID-19 infection but also do genomic tracing to understand viral spread, says Chiu.
All seven strains were extinguished by July, suggesting the early outbreak in Northern California was being driven by exogenous introduction, Chiu says. Based on that finding, he predicts “robust local control of the pandemic is only possible if there is also containment at the state and national levels.”
Chiu’s team plans to submit an Emergency Use Authorization (EUA) application to the Food and Drug Administration in the coming weeks to deploy the viral genome sequencing approach on a bigger scale, he says. The tool is currently being validated in the clinical lab of the California Department of Public Health, and the EUA will help generate data that will be needed later to support a formal marketing application either through the agency’s traditional PMA or 510(k) pathway.