Skip to main content
Premium Trial:

Request an Annual Quote

NIH Team Uses WGS to Track Lethal Bacterial Outbreak in Real Time in Hospital Setting

Premium

Researchers at the National Institutes of Health have used whole-genome sequencing to track an outbreak of drug resistant Klebsiella pneumoniae in real time, demonstrating that sequencing during an outbreak can lead to changes in patient management to stem the outbreak.

The study, which used the Roche 454 GS FLX instrument, was published today in Science Translational Medicine.

"We are now able to sequence and analyze microbial genomes fast enough to integrate that information into hospital practice to make it directly clinically relevant," Julie Segre, a senior investigator in the genetics and molecular biology branch of the NIH and senior author of the study, told Clinical Sequencing News.

"For decades, the staple tool of hospital epidemiology for tracking bacteria involved in outbreaks has been pulsed-field gel electrophoresis," added Tara Palmore, the deputy hospital epidemiologist at the NIH Clinical Center. But, it's "rather crude," she said, and can't distinguish between Klebsiella isolates. "Whole-genome sequencing just blows this tool out of the water."

This week's study adds to a growing body of evidence that sequencing can be especially useful for tracking outbreaks. The UK Health Protection Agency in conjunction with the University of Oxford had a protocol in place during the Olympics that would have allowed them to sequence pathogens on the Illumina MiSeq in the case of an outbreak during the games (CSN 8/8/2012).

The University of Oxford's Derrick Crook is also involved in the Modernizing Medical Microbiology Consortium, a research project funded by the UK Department of Health that includes the University of Oxford, the HPA, and the Wellcome Trust Sanger Institute.

The consortium recently received a four-year grant to conduct research toward implementing sequencing in the public health setting.

And, Crook's team and a separate team from the University of Cambridge earlier this summer published studies in BMJ Open and the New England Journal of Medicine, respectively, demonstrating the use of next-gen sequencing to track outbreaks in a hospital setting (CSN 6/20/2012).

The NIH outbreak occurred last June when a patient known to have a strain of Klebsiella pneumoniae that was resistant to the antibiotic carbapenem was admitted to NIH's Clinical Center. Drug-resistant K. pneumoniae is of particular concern because it has a mortality rate of 40 percent so she was immediately isolated and any staff or visitors were required to wear gloves and gowns in order to enter her room. She was discharged in July.

Then, between August and December, an additional 17 patients at the clinic became infected with carbapenem-resistant K. pneumoniae. During the course of the outbreak, 11 of the infected patients passed away, with six of the deaths attributable to the pathogen.

After patients two and three were found to be infected, the hospital used traditional typing tools to determine whether the bacteria had been transmitted to the second patient from the first or whether it was an independent event.

Neither pulsed-field gel electrophoresis nor PCR was able to definitively say whether transmission had occurred between the two patients. Both techniques identified the strain as belonging to the same lineage, however that lineage is ubiquitous in hospital settings — comprising some 70 percent of all Klebsiella outbreaks, explained Segre.

"The question we needed to answer immediately was, 'Is this a hospital transmission from patient 1 to patient 2, and were our methods of containment insufficient? Or is this possibly another patent who's come in infected and because surveillance methods aren't always perfect, maybe the previous hospital, and the patient himself, didn't know he was infected?,'" Segre said.

"That's a critical fork in the road for the hospital in terms of how they are going to respond," Segre added.

That's when Segre and her team stepped in and applied whole-genome sequencing. Using the 454 GS FLX machine, the researchers first sequenced isolates cultured from four different body sites on the original patient, revealing seven single nucleotide variants between the strains.

Meanwhile, additional patients were becoming infected, and Segre's team sequenced isolates cultured from each of the 17 other patients.

Comparing the genomes between all 18 patients revealed 41 single nucleotide variants, and also showed that the different strains clustered into two large groups and a third group containing only one patient. Cluster 1 and the patient that made up cluster 3 shared three SNVs that came from the isolates cultured from the index patient's lung and groin, while the cluster 2 isolates shared three SNVs with the isolate from the index patient's throat sample.

Analyzing the differences between the genomes allowed the team to piece together a likely transmission route in which patient 1 transmitted the bacteria to other patients on two different occasions from infections on different parts of her body.

Evan Snitkin, the lead author of the paper and a postdoctoral fellow at the NIH, said that the team used a graph-based approach, called a maximum parsimonious graph, to piece together the route.

Essentially, the analysis combines epidemiological data with genetic data to devise an outbreak map "that has the smallest number of mutations to explain it," he said.

Palmore said the results were not necessarily intuitive, and could not have been predicted without the sequence data. For instance, the order that patients were discovered to be infected did not always correspond with the order of transmission. In one cluster, for example, patient 1 most likely infected patient 3, who then infected patient 5, who infected patient 2 — despite the fact that patient 2 presented with infection 10 days earlier than patient 3.

"That was a big surprise for us," said Palmore.

Additionally, after recognizing that the pathogen had been transmitted from the first patient, despite the level of precaution that was taken, the hospital realized that further steps needed to be implemented.

The analysis of the sequence data made it "very clear that it was patient-to-patient transmission," Segre said. "The contact isolation that had been used for patient one was insufficient, so we needed to step up hospital control."

So, while the sequencing didn't change the care of the infected patients, it did change how the hospital managed the outbreak. "By showing that patients 2, 3, and 4 are linked to patient 1, that supported the hospital's decision to open up a cohorted area," Segre said.

Instead of just putting the infected patients in contact isolation, as additional patients became infected they were put into their own section of the hospital. Any staff that was caring for those patients would care only for those patients and equipment would not be transferred between the cohort area and the rest of the hospital.

The research team also tested various pieces of equipment for the pathogen, and in one case detected the pathogen on a ventilator that already been extensively cleaned, including with bleach. The ventilator had only been used on an infected patient, so was not responsible for any transmission events, but finding it "demonstrated that the organism could be tenacious in the environment, and could evade standard cleaning techniques," Palmore said.

The pathogen was also found in sink drains and on other equipment, but transmission from the environment to a patient was never demonstrated, Palmore said.

For this study, sequencing was done on the 454, and at the time, the NIH sequencing center had only that instrument and an Illumina HiSeq. Now, the center also has two MiSeqs and is purchasing a third.

Segre noted that because turnaround time is so important, in future outbreaks, sequencing would likely be done on the MiSeq.

However, she said that it is still important to have reference genomes sequenced on multiple platforms because it adds an extra level of confidence for distinguishing between true variation and sequencing error.

Segre thinks that sequencing will begin to be used more routinely for tracking outbreaks, and Palmore said that she would definitely want the technology to be used in future outbreaks.

"After seeing how powerful whole-genome sequencing is in the setting of an outbreak and elucidating hospital epidemiology, it's hard to imagine I'll be satisfied going back to our traditional tools," Palmore said.