Skip to main content
Premium Trial:

Request an Annual Quote

Metagenomics Shows Potential for Foodborne Pathogen Outbreak Investigations


NEW YORK (GenomeWeb) – Culture-independent diagnostic tests (CIDTs) such as antigen-based assays and molecular methods promise faster and more accurate bacterial pathogen detection than traditional culture, and are particularly valuable in outbreak scenarios where the source of the infection is often unknown.

A recent study conducted by researchers from the US Centers for Disease Control and Prevention and Georgia Institute of Technology sought to test the strengths and weaknesses of one such CIDT, whole-genome shotgun metagenomics, for detecting food-borne pathogens and signs of co-infection that traditional methods might miss.

For the study, published last month in Applied and Environmental Microbiology, the researchers analyzed stool samples gleaned from two geographically isolated food-borne outbreaks where the causative strains were known to be distinct strains of Salmonella enterica based on cultures performed on the samples. According to the paper, the results obtained using shotgun metagenomics were consistent with those obtained using cultures. Furthermore, the researchers were also able to assess pathogen abundance and diversity in the samples, the potential for coinfections with other bacteria, and changes to the healthy gut microbiome as a result of infection.

The researchers claim that theirs is one of the first studies to apply a metagenomics-approach to pathogen detection in acute diarrheal cases. The study brought together researchers from Georgia Tech's School of Civil and Environmental Engineering and the CDC's Enteric Diseases Laboratory Branch, which manages PulseNet, a national network of laboratories that focuses on solving outbreaks of foodborne disease

One of the paper's authors, Andrew Huang, is a microbiologist and bioinformatician at the CDC and co-leads a group there that is working on culture-independent and metagenomics subtyping. He explained to GenomeWeb that his team began exploring metagenomics for foodborne infections because of the ongoing evolution in pathogen detection technology.

Lower cost PCR-based tests are rapidly becoming the predominant method for testing pathogens, replacing older techniques that required researchers to isolate bacteria from samples. With PCR-based testing "we are not going to have those isolates anymore which is what we used to use to get these DNA fingerprints [that] say these were the samples that were involved in an outbreak," Huang said. Metagenomics offer the possibility of being able to do that kind of same fingerprinting that culture-based methods offer directly out of a stool sample. It is also "more reliable and robust" than antigen- or PCR-based approaches, which suffer from technical limitations such as amplification bias, the researchers wrote.

Furthermore, metagenomics offers more fine-grained detail about the microbial domain than the standard culture-based techniques that have been the mainstay of infectious disease diagnostic testing for decades. According the paper, "a substantial fraction of the gut microbial community, including opportunistic and rare pathogenic microbes, remains uncultivated and hence undetectable by culture-based tests." As a result, more than 38 million cases of foodborne illness each year cannot be attributed to specific causes using these methods.

Enteric foodborne diarrhea made an ideal test case for a metagenomics-based approach because in many cases "the agent frequently remains unidentified either due to technical challenges or the fact that several diarrheal infections quickly self-resolve," the researchers wrote. 

The outbreaks in question, which happened in Alabama and Colorado, were particularly useful because of their timing, according to Huang. In this case, both outbreaks happened in the same month which meant that researchers could assess whether they stemmed from a common source — their analysis showed they did not. 

Also, infected patients became ill very quickly — in as little as two to three hours — suggesting that they likely had higher pathogen loads which would provide a much stronger signal for testing. "One of our big problems is that there's a lot of bacteria in stool so its like looking for a needle in a haystack when you are looking for salmonella amongst all the other bacteria," Huang said. A stronger signal also made it possible to tell the difference between the closely-related samples, he added.

Other questions of interest to the researchers were the effects of infection on a healthy microbiome, according to Kostas Konstantinidis, an associate professor of environmental engineering and sustainable communities. "We don't fully understand how some of these pathogens are causing disease," he told GenomeWeb. "Whether and to what extent they alter the natural healthy gut microbiome is a question that can give us some hints about how exactly they are causing the disease." Assessing possible co-infections offers a more well-rounded and unbiased picture of the patient's disease and could also explain disease severity. 

In the study, the researchers obtained 11 stool samples from the public health labs in both outbreak regions, extracted the DNA using the QIAmp DNA minikit from Qiagen and prepped it for sequencing with the Illumina NexteraXT sample preparation kit. They then sequenced the samples on an Illumina MiSeq instrument. For the metagenomic assembly, researchers used the IDBA-UD software. They were able to identify a Salmonella Heidelberg strain as the cause of the diarrhea observed in the patients — this matched results from independent testing using cultures. They also confirmed that the outbreaks were caused by highly related but distinct Salmonella strains which was consistent with the findings from previous testing. Their results also indicated possible co-infection with Escherichia coli and Staphylococcus aureus in some of the patients.

Both Konstantinidis and Huang believe that their work could play a valuable role in future efforts to develop sequence-based diagnostics and subtyping methods for pathogens in enteric diarrheal cases as well as other clinical infections. "We are really excited about this because this could really change the way that we detect and monitor food-borne disease," Huang said. "Because it's faster, it could really help us get to that source more quickly and consequently save more lives."

However, there is still work to be done before the method can be routinely used. Huang and Konstantinidis said that cost will be a key consideration in the development of any diagnostic. That means instrument vendors will need to develop more cost-effective sequencers that can generate data more cheaply. "It's still expensive so right now I think it can be used only in special cases where traditional methods cannot tell [us] what is going on or [where] there are life-threatening outbreaks that we need to get high resolution," Konstantinidis told GenomeWeb. There are also challenges associated with determining the best protocols for collecting high-quality clinical samples and preserving them during transportation to avoid DNA degradation.

The community will also need to standardize the methods used for detection and provide training to help bring lab personnel up to speed on the technology. "It took us months to get here because we didn't know how to do it," Konstantinidis said. For example, the researchers had to come up with a method for estimating how much of the DNA sample they had characterized. They also had to figure out ways of removing human DNA from samples as well as for identifying the pathogen signal in samples. 

"Every one of these steps required us to sit back and try to come up with a best approach to do it," he said. "It is a multistep process and every step took a lot of time and meetings. Their efforts do appear to have paid off because the team expects that it will soon be able to complete both the sequencing and analysis within a day.

Besides diagnostics, there are also applications on the basic science side that could benefit from metagenomics approaches. For example, in his lab, Konstantinidis is studying the effects of bacterial infections on the healthy gut microbiome including whether different pathogens have different effects because they express different toxins or have different physiology. "[We] need to understand better how different pathogens work and how they mess up the gut microbiome and also what kind of therapeutic treatments will come out of this," he said.

For their next steps, the researchers plan to re-run their analysis on more typical stool samples. The samples in the recently published study came from atypical outbreaks with patients having high quantities of the salmonella pathogen in their samples and developing diarrhea rapidly, Huang explained. They will also use the method to study cases of E.coli infection.

Moving forward, the researchers plan to test whether the approach works as effectively with samples from patients with lower salmonella levels who developed diarrhea more slowly, he said. Konstantinidis said that the partners plan to publish another paper that will provide details on what lower detection limits for bacterial pathogens in samples are. The researchers also plan to compare the samples from sick patients to healthy gut microbiomes to see where the differences lie.

Other plans include trying to improve different aspects of the sequencing and analysis workflows. On the bioinformatics side, there are faster ways to do the analysis, Konstantinidis said. For example, they are exploring whether or not they can get the same resolution from kmers than they do from looking at the full DNA sequence, he said. On the wet lab side, they are exploring new methods for removing the human DNA from samples prior to sequencing. They also hope instrument vendors continue to improve their sequencing platforms as well as create more efficient methods for DNA extraction and library preparation, he said.

They also hope to identify agent-specific signatures that could be used to detect the pathogens in future outbreaks. That way, "we don't have to go through the lab and isolate organisms, which takes a couple of days," Konstantinidis said. "We can do it pretty quickly by looking at the signature." They will, however, need to test many more samples from more outbreaks before they can make any conclusions about pathogen-specific signatures for diagnostic purposes.

On the whole, these early results look "very promising," Konstantinidis said. "Not only were we able to get results that were of similar quality and resolution as the traditional methods but we were all able to get results that were not easily attainable by the traditional methods. The potential to optimize the technique so you can get an answer within a day is great."