When a patient develops signs of a bacterial infection, clinicians often send a specimen to a laboratory to see if the illness-causing culprit can be identified. This process typically involves trying to coax bacteria from the specimen to grow in culture—and because this process can take days and doesn’t always work, clinicians have long sought a faster and more efficient alternative. Results from a study published in JAMA today suggest that genomic sequencing may one day be such a tool.
In a proof-of-principle study of the use of a “metagenomic” sequencing technique in diagnosing bacterial infections, the scientists collected samples taken from 45 patients who developed diarrhea during an outbreak of Shiga-toxigenic Escherichia coli (STEC) in Germany in 2011, 40 of whom had laboratory-confirmed STEC and 5 of whom developed diarrhea unrelated to the outbreak. The researchers demonstrated that the approach was able to identify the outbreak strain in 27 of 40 samples from patients who were known to be positive for STEC. In addition to identifying additional unexpected (non-STEC) strains of bacteria in samples from the non–STEC-infected patients, further testing revealed Clostridium difficile in samples from some STEC-positive patients, raising questions about whether these bacteria might also be contributing to illness in these patients.
Mark J. Pallen, MA, MD, PhD, of the University of Warwick in the United Kingdom, discussed the findings with news@JAMA.
news@JAMA: Can you briefly describe metagenomics?
Dr Pallen: It is where you extract DNA from a sample, typically one that represents a complex microbial community, and do “shotgun” sequencing to see what you can deduce. It’s a brute-force approach. The advantage is that it is open-ended—you can find things you weren’t looking for, the “unknown unknowns.”
In 2004, Craig Venter and his colleagues published a landmark metagenomics paper looking at the organisms in the Sargasso Sea. Developments in technology are rapidly making this approach more useful. You can now achieve a depth of coverage that would have been unthinkable years ago.
news@JAMA: What advantages might metagenomics have over bacteria culture?
Dr Pallen: If you go into a clinical diagnostics lab and are relying on culture, there are different requirements for each type of pathogen. Many pathogens will grow overnight, some might require days or weeks, and some won’t grow at all. The promise of metagenomics is that you can one day replace those work flows with a one-size-fits-all work flow. The complexity with metagenomics comes with analyzing the data. We are not there yet, but our results represent a proof of principle. Making this process more user friendly will follow.
news@JAMA: How were you able to zero in on the outbreak strain in such a bacteria-rich environment, which also would include E coli strains that one would normally find in healthy people?
Dr Pallen: We adopted 2 approaches. First, knowing the outbreak strain, we took a look at DNA in the samples to see if they matched the outbreak strain—and yes, we could.
We then took a prospective approach. We made the assumption that the outbreak strain had to be present in half the samples from the outbreak and excluded sequences we would expect to find in normal people [based on] data from the European Metagenomics of the Human Intestinal Tract (MetaHit). That was enough to come up with a series of sequences from the outbreak strain. E coli strains are very variable but all have a core genome, which we would have excluded along with other sequences we would expect to find in healthy people. Once we got the outbreak-related sequences, we could then go back and pull back the core sequences and reconstruct a near-complete genome of the outbreak strain.
It’s a bit like doing a jigsaw puzzle. You can look at the picture on the box and figure out how to put it together; that’s what we did the first time. Our alternative approach was to do the puzzle without any prior knowledge.
news@JAMA: Why didn’t you find the outbreak bacteria in 13 of the samples from people known to be infected with STEC?
Dr Pallen: We asked our German collaborators to give us a selection of stool samples: some in which the outbreak strain was highly abundant, some where it was moderately abundant, and some where it was only in the smallest amounts. We had samples with very, very low bacteria loads, and those were the ones that didn’t give us any results. I’m not too upset about the fact that we didn’t get in hard cases. In each of those samples, we know the outbreak sequences are there, but we didn’t go deep enough with the sequencing to find them. It’s a failing of current sequencing practice. I’m optimistic that we are going to get deeper and deeper sequencing abilities.
news@JAMA: What do you make of finding unexpected strains of bacteria in the samples?
Dr Pallen: We found a few sequences we didn’t expect. Campylobacter concisus is an emerging cause of diarrhea. We found C difficile in some of the cases we knew had STEC. We don’t know if it was simply proliferating in the face of the inflammation or if it was making a contribution to the disease.
news@JAMA: What do you think is the main take-home message?
Dr Pallen: We’ve achieved proof of principle that culture-independent sequencing can come close to what you can achieve by culturing easy-to-grow bacteria. We accept that it is not quite ready out of the box. But I think in 3 or 5 years’ time, this technology will become more applicable to the everyday clinical environment.