Skip to main content
Premium Trial:

Request an Annual Quote

Seven-Year Study Validates Metagenomic Sequencing for Diagnosing CNS Infections

Premium

NEW YORK – New research suggests metagenomic sequencing of cerebrospinal fluid (CSF) samples can help in diagnosing bacterial, viral, fungal, or parasitic infections of the central nervous system (CNS) and has better sensitivity than other testing methods.

"Unlike other diagnostic tests, metagenomic NGS testing is an agnostic, untargeted approach to diagnosis that does not require clinical suspicion of the cause of infection a priori," senior and corresponding Charles Chiu, a laboratory medicine and infectious diseases researcher at the University of California San Francisco and codirector of the UCSF Clinical Microbiology Laboratory, said in an email.

Chiu and his colleagues developed clinical metagenomic sequencing for infectious disease diagnostics about 10 years ago and launched their test at seven participating hospitals in 2017.

As they reported in Nature Medicine on Tuesday, they performed metagenomic next-generation sequencing (mNGS) on 4,828 CSF samples collected from individuals with apparent CNS infections in the US and beyond from June of 2016 to April 2023.

"Here, we sought to evaluate the clinical applicability of mNGS testing performed over seven years as part of the diagnostic workup for a geographically broad population of patients with suspected but 'difficult-to-diagnose' CNS infection," the authors explained.

While some 2,420 CSF samples came from patients in California, they also assessed samples from 45 other US states and from sites in Canada, Australia, Sweden, Brazil, Latvia, Colombia, and Portugal.

By combining the metagenomic sequence profiles with computational analyses and clinical data, when available, the team found evidence for CNS infection in 14.4 percent of patients tested, reaching accurate diagnoses in 86 percent of them.

"The test was able to identify a broad array of pathogens that are difficult to detect using conventional methods, including novel, emerging, and/or unexpected microorganisms that were not considered and tested a priori," the authors reported.

In contrast to targeted sequential testing, the mNGS approach provided a boost in sensitivity and accuracy when it came to diagnosing CNS infections such as encephalitis, meningitis, or meningoencephalitis in hospitalized patients.

Overall, the mNGS CSF testing approach was 92.9 percent accurate, with 99.6 percent specificity and a sensitivity of 63.1 percent. That sensitivity rose to 86 percent when the investigators only considered cases diagnosed by CSF direct detection testing. In contrast, they found that conventional direct detection testing methods that focused on CSF had 45.9 percent sensitivity and those focused on non-CSF had 15 percent sensitivity, while indirect serologic testing had 28.8 percent sensitivity.

"Our results indicate that CSF mNGS testing should be a routine part of our clinical armamentarium in the future, as 48 (21.8 percent) of 220 [diagnosed] infections from the UCSF cohort were detectable only by mNGS," Chiu noted. "This is especially true for hospitalized patients with critical neurologic illness, with many patients in the study undergoing long and extensive diagnostic workups that can take weeks to months without yielding a diagnosis."

Even so, the team found that the test appeared to be most accurate when combined with clinical data and feedback from treating physicians, since a slightly higher proportion of California-based cases was diagnosed than cases from other parts of the US or beyond.

Together, the authors suggested, the latest findings "justify the routine use of diagnostic mNGS testing for hospitalized patients with suspected CNS infection."

With that in mind, Chiu and colleagues, along with Joe DeRisi of the Chan Zuckerberg BioHub, Pardis Sabeti of Harvard University, and others, cofounded a startup called Delve Bio in 2023 that has licensed the technology from UCSF in an effort to expand clinical access to the mNGS CSF test.

The investigators are also looking into pairing metagenomic sequence data with artificial intelligence to track human host immune responses to infection or other conditions such as autoimmune disease.

"For our next steps, we are focused on investigating future application of mNGS testing, particularly in characterizing RNA host response profiles from human host mNGS data," Chiu said, adding that "we now have well-pedigreed samples with clinical metadata from thousands of patients and collected over seven years to develop, validate, and deploy these host-response models."

In a related paper published in Nature Communications, meanwhile, members of the same team looked at the feasibility and performance of mNGS for agnostically identifying pathogens causing respiratory illnesses such as pneumonia or COVID-19 using upper respiratory swab or bronchoalveolar lavage samples.

Though the authors noted that the cost of their sequencing approach still exceeds that of conventional respiratory viral multiplex panel testing, they suggested that "the benefits for mNGS testing of greatly expanded scope of detection, capability to identify novel emerging viruses, and comparable performance likely outweigh the costs under certain clinical and public health scenarios."