How a CNS Infection was Diagnosed in 48 Hours
ARTICLE IN BRIEF
Less than 48 hours after the serum and CSF samples were fed into a new-generation sequencing machine, a research team was able to identify and treat a previously undiagnosed illness, neuroleptospirosis.
A rapidly progressing case of meningoencephalitis in a 14-year-old boy was diagnosed as neuroleptospirosis in 48 hours with a new, relatively inexpensive genetic sequencing system that, neurologists say, promises to transform the way such cases are routinely handled.
Although only a handful of academic centers presently have access to both the rapid sequencing technology and the bioinformatics capabilities necessary to decipher its output, a wider rollout is expected over the next few years. In the meantime, the neurologist who served as first author of the new report told Neurology Today that clinicians faced with similarly serious, unknown CNS infections should contact his team.
“We can accept samples from external folks,” said Michael R. Wilson, MD, an assistant professor of neurology at the University of California, San Francisco (UCSF), where he works in the genomics laboratory of study coauthor Joseph DeRisi, PhD. “We're very interested in seeing more cases like this.” (Interested neurologists may e-mail him, he said, at [email protected].)
UNRAVELING THE CASE
As reported first in the June 4 online edition of New England Journal of Medicine, the patient, under the care of an immunologist for severe combined immunodeficiency, presented three times to a Wisconsin medical facility over the course of four months “with fever and headache that progressed to hydrocephalus and status epilepticus necessitating a medically induced coma.”
The cause of the boy's illness — as is evident in more than half the cases of meningoencephalitis — remained undiagnosed despite extensive clinical testing, including MRI, and biopsy of the right frontal lobe. In addition to his routine monthly IV immune globulin therapy, intravenous glucocorticoids were administered for possible neurosarcoidosis, but his clinical status only worsened.
UCSF researchers were asked to help diagnose the case by a pediatrician at the University of Wisconsin School of Medicine, James Gern, MD, who was collaborating with them on an unrelated study. Serum and CSF were sent overnight, and then fed into an Illumina MiSeq instrument. One of the latest iterations of so-called “next generation” sequencing devices, it generated 8,187,737 sequences. Then, in just 96 minutes, those sequences were analyzed with a bioinformatics pipeline developed at UCSF, comparing them with reference sequences in the National Center for Biotechnology Information databases of bacteria, viruses, fungi, and parasites.
In all, less than 48 hours after the samples had reached UCSF, 475 sequences corresponding to the leptospira genome were identified in the patient's CSF but not in the serum, with no other convincing hits to other bacteria or viruses.
Without waiting for the identification to be confirmed by targeted PCR — performed soon after by two independent laboratories — the Wisconsin team decided to begin treatment for neuroleptospirosis. Glucocorticoids were tapered, and antibiotic coverage was narrowed to high-dose IV penicillin. The patient recovered over the next seven days, and was discharged to inpatient rehabilitation another 14 days later. He returned home after 11 days of rehabilitation close to his premorbid functional status.
“The old tools and old approaches had pushed the clinicians toward a wrong answer,” said Kenneth L. Tyler, MD, FAAN, who specializes in CNS infections as chair of neurology at the University of Colorado School of Medicine and is an associate editor of Neurology Today. “They understandably but incorrectly suspected neurosarcoidosis and were using the wrong therapy, and it could have had negative consequences.”
Although the term “next generation” sequencing has been applied to other high-throughout devices since 2007, “the unique thing in this case is that they could use it in real time, so that it became clinically applicable,” said Avindra Nath, MD, the intramural clinical director of the National Institute of Neurological Disorders and Stroke and a specialist in neuroimmunology.
“The technology is getting better and better and better,” he said. “But before it can become a standard diagnostic tool,” he added, “ultimately it has to become automated, reproducible, and idiot-proof. That way LabCorp can do it for you.”
Dr. Wilson said he is convinced that the technology will be adopted as a routine tool at major medical centers in coming years. “It's been breathtaking how rapidly it's advanced,” he said. “A few years ago, the same result we achieved in hours would have taken a month.”
W. Ian Lipkin, MD, professor of neurology at Columbia University's College of Physicians and Surgeons, published a paper in 2008 using what was then state-of-the-art high-throughput, unbiased sequencing to identify a novel arenavirus as the cause of a febrile illness with varying degrees of encephalopathy in three transplant patients. Those patients, however, had died by the time the diagnosis was made.
The new paper, Dr. Lipkin told Neurology Today in an e-mail, “reinforces the need in clinical microbiology for rapid, sensitive molecular diagnostics.”
The sequencing device is about the size of a small television — 22” by 22” by 20” — and is said to cost about $125,000, although prices are not listed on the company's website.
“Illumina is of course trying very hard to come up with a less and less expensive, more and more modular and easy-to-use machine, so it isn't that much harder or more expensive than other automated machines used all the time in hospital diagnostic laboratories,” Dr. Tyler said. “Right now their webinars and advertising make it sound like it's easier than recording ‘Game of Thrones.’ It isn't, but it's getting there.”
A particular benefit of the unbiased sequencing, he said, is that it doesn't require physicians to tell their laboratory which pathogens to test for.
“You currently have to specify tests virtually virus by virus,” he said, “which is moronic and leads to all sorts of errors — things that you forgot to list, or things you never should have listed. This new approach actually looks for all the possibilities. They get these millions of sequences, but the beauty of the bioinformatics is that it allows them to get down from 8 million to the actual culprit.”
Even then, though, “You're only as good as the databases you have,” Dr. Tyler said. “If something isn't in the database that you search, you won't identify it with even these new devices.”
EXPERTS: ON SEQUENCING FOR NEUROLEPTOSPIROSIS