Kristine Wylie and Todd Wylie look at genomic data on their computer screen. Similar data were used to diagnose a difficult and unusual case of chronic meningitis.
A patient who suffered for three years from confusion and difficulty walking was successfully treated after a research team at Washington University School of Medicine in St. Louis used an experimental technique to identify the cause of his symptoms.
The researchers sequenced tens of millions of bits of genetic material in the patient’s cerebrospinal fluid using a technique known as high-throughput metagenomic shotgun sequencing. They found that an acne-causing bacterium that normally lives harmlessly on the skin had infected the tissue surrounding his brain and spinal cord, causing chronic meningitis.
The paper is available online in the journal Transplant Infectious Diseases.
John DiPersio, MD, PhD, deputy director of the Siteman Cancer Center, Armin Ghobadi, MD, an assistant professor of oncology, and colleagues already were treating a 40-year-old patient for leukemia when he developed a persistent headache, confusion and difficulty walking. The doctors suspected an infection, but traditional microbiologic techniques failed to identify the pathogen, and multiple courses of antimicrobial drugs failed to permanently eliminate the symptoms.
In desperation, the doctors reached out to colleagues at the School of Medicine and the university’s McDonnell Genome Institute. Through work on the Human Microbiome Project and other genomics research, these scientists had developed the expertise and scientific infrastructure necessary to identify unknown microbes that are undetectable by traditional methods.
The task fell to infectious disease specialist Gregory Storch, MD, the Ruth L. Siteman Professor of Pediatrics, and Kristine Wylie, PhD, and Todd Wylie, assistant professors in pediatrics.
Using two samples of cerebrospinal fluid drawn from the patient three months apart, the researchers sequenced the genetic material and then compared the sequences to a database of known genomes to identify the organisms present.
“Lo and behold, Propionibacterium acnes dropped out,” said Storch, referring to a skin bacterium mostly known for causing acne. In both samples, about 47,000 of the millions of sequences belonged to P. acnes, five times as many as the second most common microbe.
The researchers also analyzed five control samples from other patients, two of whom had viral infections and three whose infections were unknown. The method correctly identified viral sequences in the two samples from known viral infections. Just a few hundred P. acnes sequences were found in the control samples, compared with the tens of thousands seen in the sample from the patient with meningitis.
“Meningitis isn’t often caused by P. acnes, but in this case, it fit the clinical picture,” Storch said. “We knew we had to have an organism of low virulence, meaning it was capable of causing disease but not severe disease that would be rapidly fatal.”
There are rare reports in medical literature of P. acnes meningitis. The cases usually involve immunocompromised patients similar to DiPersio’s and do not progress rapidly. The bacteria had been cultured from a previous sample of the patient’s, but it had been dismissed as an insignificant contaminant from the skin.
Since the technique is experimental, the researchers had to obtain permission from the Food and Drug Administration and the Washington University Institutional Review Board, which oversees studies involving human participants, before their results could be used to determine treatment for the patient.
DiPersio and a team including Merilda Blanco-Guzman, MD, and Steven Lawrence, MD, co-authors of the published report, administered a six-week course of antibiotics targeted against P. acnes. The antibiotic cleared the infection, and, after fluid that had collected in the patient’s central nervous system was drained, his neurological symptoms resolved.
Metagenomic shotgun sequencing was originally developed to identify the members of microbial communities in the environment, such as those found in soil or seawater. This study is only the third to report using this technique for clinical diagnosis.
“What’s novel about this case is that we were able to identify an organism that would otherwise be overlooked because it is not normally a pathogen, and that led to successful treatment of a patient who had a very complicated history,” Kristine Wylie said.
Current methods for diagnosing infection require technicians to grow the organism or amplify its genetic material, both of which require some knowledge of what the organism is likely to be. Shotgun sequencing requires only that the genetic material be present.
“Even an unknown organism could be identified, if it had some relationship to something in the database of microbial sequences,” Storch said. “We might be able to identify it as, say, a coronavirus, even though it might be a new species of coronavirus.”
The researchers currently are working to validate the technique in clinical cases. After the first successful diagnosis, they tried the technique on another difficult case. That time, it didn’t work. The cause of that patient’s illness is still unknown.
“We didn’t detect anything,” Kristine Wylie said. “It could be the sample was taken at the wrong time, and the infection already was cleared. Or it might not have been an infection at all.”
The researchers have begun another study using samples from 20 difficult cases of suspected infections that resolved without the specific cause ever being identified.
The researchers want to gain insight into which kinds of cases would be the best candidates for this technique. Compared with current methods, it is slow and expensive. The meningitis case took five days to diagnose, at a cost of about $7,000.
“It’s expensive, but these difficult cases are already very expensive,” said Kristine Wylie. “In the case of this patient, he was admitted to the hospital multiple times before he was diagnosed.”
Storch agreed. “We see this as something held in reserve for very difficult cases,” he said. “It was clearly useful in this one case, but we need to get a much better handle on what kinds of patients and what kinds of diseases it would be worth trying this on.”