NEW YORK – Researchers from Stanford University and the University of California, San Francisco have used a metagenomic next-generation sequencing test originally developed to detect neurological infections to uncover neoplasms of the central nervous system.
This application broadens the test's scope and advances the research team's plans of eventually commercializing a simple test that can provide clinicians with a complete report of sequence-based health data, such as infectious pathogens, cancer and cancer risk, and various types of host response.
Wei Gu, Charles Chiu, and their colleagues tested their mNGS assay in two case-control studies published together on Sep. 13 in JAMA Neurology. In the first, they tested the assay's performance in a set of CSF samples from patients with known CNS cancers or other neurological disorders.
In the second, the team evaluated patients with symptoms suggestive of neuroinflammatory disease but who were later diagnosed with malignant CNS tumors, representing patients whose tumors likely could have been detected earlier with the less-invasive mNGS assay.
All CSF samples had been previously obtained in other studies, meaning that patients needed to undergo no new tests.
"The ability to reanalyze the same data and get clinically useful information, I think, is a really efficient way to enhance the utility of an existing test," Gu, assistant professor of pathology at Stanford and the study's first author, said in an interview.
Specifically, the researchers used their test to look for large copy number variations, or CNVs, indicative of cancer, such as aneuploidy — the presence of an improper number of chromosomes stemming from either their loss or conversely, from gaining extra copies.
Aneuploidy and other large CNVs occur in an estimated 90 percent of all malignant tumors and are prevalent in both primary and metastatic CNS tumors.
Overall, the mNGS method showed a 75 percent sensitivity and a 100 percent specificity, detecting aneuploidy in 64 percent of the patients in the test performance study and 55 percent of those in the neuroinflammation study, who were later diagnosed with CNS cancers.
Most of the patients in whom the researchers detected aneuploidy had multiple copy number variations with tumor fractions between 31 and 49 percent.
In several cases, imaging and cytologic tests suggested benign lesions, which brain biopsies later confirmed as malignant. Evidence from the mNGS test often corroborated biopsy results.
Having a fast, reliable, and minimally invasive method to detect and identify CNS tumors could meaningfully impact treatment decisions and health outcomes, such as deciding whether or not to perform surgery, depending on whether a patient has a CNS lymphoma versus a glioma.
Adapting the mNGS assay to test for CNS cancers grew out of the same group's prior work involving using it to diagnose challenging cases of neurological infections.
That study showed that essentially scooping up and analyzing all DNA and RNA in a CSF sample revealed the presence of pathogens that might be missed in other tests in a highly unbiased manner. Chiu refers to this as a "hypothesis-free" approach to diagnosis, in which physicians use a single test to explore multiple possible diagnoses.
Despite the ability to detect a wide range of pathogens and malignancies, the test will still likely be used alongside others, rather than as a standalone.
"I don't anticipate that the test is going to replace existing tests," Chiu said, "but it's very complimentary in the sense that you can sort of identify pathogens that would otherwise not be identified by conventional testing."
No conventional tests exist, for instance, for numerous fungi, bacteria, and other pathogens. Nonetheless, the mNGS assay sometimes fails to find some pathogens identified through serological testing.
Finding tumor DNA in the CSF also relies to an extent on a lesion's proximity to that compartment. The farther a tumor is from having direct access to CSF, the less DNA it sheds into that space.
UCSF currently runs the pathogen assay as a lab-developed test, which received US Food and Drug Administration breakthrough device designation in 2019. The lab does not currently run client samples for clinical CNS cancer detection, although Chiu commented that they are discussing doing so in the future.
Chiu and his colleagues hope to obtain full regulatory approval in the next six to 12 months. If successful, approval would mark a unique achievement.
"This is fundamentally a new type of test," Chiu explained, "and the FDA to my knowledge has never approved a metagenomic test that's this broad for clinical use."
The group is now working to lower the test's cost and speed.
The test currently costs roughly $2,400 to run, although this is expected to drop significantly as it is scaled up from running slightly under 20 samples per run to potentially hundreds. The turnaround time should similarly fall to a target of 24 to 48 hours (down from closer to a week now), as the assay transitions from running on an Illumina platform to an Oxford Nanopore Technologies platform. Even shorter turnaround times will be possible as the test eventually becomes available to other labs to run on their own.
In contrast to Illumina sequencing, which relies on reversible terminators to read out sequences, nanopore technology involves a DNA molecule passing through a protein nanopore and triggering a change in the protein's current. DNA sequence is decoded in real time from the way the DNA molecule's shape, size, and length affect the change in current.
Gu, Chiu, and several colleagues demonstrated the combination of nanopore sequencing with the mNGS assay in a paper published earlier this year in Nature Medicine. Testing for infectious pathogens in a variety of patient body fluids, the team exhibited complete turnaround times of approximately six hours, including a median 50-minute sequencing run.
The UCSF team continues to actively develop the metagenomic detection test, aiming to make it more universal in scope. They are also evaluating potential commercial partners, who can help refine and clinically validate the test.
"I think down the line, you will see more and more of these tests," Gu said, "especially for patients without a diagnosis. More tests that can look at the data in different ways and be able to cover multiple diseases."
Chiu reiterated that the test's new use in detecting malignancies required no change in the test itself, nor the development of an entire new test protocol.
The bioinformatic basis of the test also means that the more widely it gets deployed, as would occur with regulatory approval, for instance, the more data becomes available for future analyses, including of past patient samples.
Chiu anticipates that the test will eventually result in a comprehensive report covering nonhuman reads related to infections, as well as human reads for cancer prediction and analysis, and potentially even RNA reads related to host response.
The team is currently working on a study, to be published in the near future, in which this same test is applied to distinguishing different host responses, such as autoimmune disease, viral, and bacterial infections.
"There's a lot of work and resources and effort that is really required to get these tests through regulatory approval, but if we can get just the wet lab protocols approved by the FDA, then to get approval for these secondary analyses simply requires getting approval for the bioinformatics analysis," Chiu said. "That is still challenging, but it's certainly far less challenging than having to develop a test from scratch."