For several reasons, 1992-93 was a watershed period in the annals of pathogenic food contamination in the U.S. That’s when an outbreak involving the up-and-coming bacterial mutant known as E. coli O157:H7 entered the public consciousness with the Jack in the Box poisonings.
In terms of fatalities, Jack in the Box wasn’t overwhelming, though the gruesomeness of the deaths of its youngest victims was wrenching. As a public health event, however, the calamity was a call to action.
More than 700 people were sickened before public health investigators recognized the hamburger common thread between the cases. A 39-day delay in responding to a foodborne outbreak was a clarion call, and in 1996 the Centers for Disease Control (CDC) in partnership with the Assn. of Public Health Laboratories launched PulseNet, the molecular surveillance and database of infectious disease outbreaks.
If investigators a quarter of a century ago were supported by whole genome sequencing (WGS) technology, would it have taken 39 days to recognize the hamburger connection? “Probably not,” offers Peter Gerner-Smidt, chief of the CDC’s Enteric Diseases Laboratory Branch, Atlanta, the backbone of PulseNet. “I don’t know how far down we can get” in terms of identifying an outbreak, given the need for epidemiologists to interview patients and detect patterns that suggest an outbreak and its source.
That said, outbreaks are being identified earlier and when they involve fewer victims. The 83 public health and regulatory labs affiliated with PulseNet are in the process of transitioning from pulsed field gel electrophoresis (PFGE), the go-to sample screening method for PulseNet’s first 20 years, to WGS, a much more precise method of subtyping bacteria’s DNA and allowing microbiologists to characterize the serotypes of species.
Without apology to J.K. Rowling, Gerner-Smidt likens PFGE to reading the chapter titles of a Harry Potter novel, while WGS is analogous to reading the text.Technicians at public health labs in 46 states will have been trained in WGS subtyping from specimens by year end, and CDC is in the final stages of validating the analytical methods for Listeria, Campylobacter, Clostridium botulinum and other pathogens to follow.
“At some point, the clinical labs will sequence and analyze and do the comparisons” of the molecular fingerprints, the microbiologist predicts. “That’s the brave new world that’s probably 10 years away.” That would collapse the ability to match the source of multiple illnesses from weeks to days.
One speed bump to that new world is the adoption of multiplexed molecular panels by clinical labs to test patient stool samples. While those panels greatly accelerate and improve patient treatment, they pre-empt the need to set isolates aside for surveillance, depriving PulseNet labs of the ability to catalogue DNA sequences. The solution, Gerner-Smidt suggests, may be “affordable metagenomic sequencing-based pathogen detection and subtyping tools.”
FSMA inspections contribute
When PulseNet marked its 20th anniversary in 2016, its pathogen library included close to 900,000 DNA fingerprints, with 2015’s 89,000 samples the high water mark. Today, the database is approaching 1.1 million, augmented in part by environmental samples collected during FDA inspections to enforce FSMA preventive controls rules.
Given how quickly bacteria evolve and even mutate, it would be extremely difficult to match a pathogen collected from a drain in a food plant in 2017 to foodborne illnesses in 2027. On the other hand, there have been Salmonella outbreaks 10 years apart where there were scarcely any differences in the strains. “We also see them change quite a lot,” he cautions. “It depends on how they propogate.
“When we have sequencing matches in the database, pretty often they originate from the same source,” continues Gerner-Smidt, “but still we must interview the patients” to establish a link. A listeriosis event in 2013 involving farm stand cheese appeared to be related to poisonings three years earlier, but investigators were unable to interview the earlier patients and couldn’t tie the events to a common source.
Fast-forward two years to the Listeria monocytogenes contamination involving ice cream produced at two Blue Bell Creameries. Routine sampling by state health officials in South Carolina isolated Listeria in single-serve products produced at Blue Bell’s Brenham, Texas, facility. Additional product samples and swabs from freezer tunnel drains in the facility produced seven different PFGE patterns that were uploaded to PulseNet.
Four of those patterns matched five patients at a Kansas hospital over a 12-month period ending January 2015. WGS highly related four of those isolates. Additional sampling implicated Blue Bell’s Broken Arrow, Okla., plants, and WGS determined it was a different strain. Two months after the investigation began, the PulseNet database was able to tie listeriosis cases in Arizona, Oklahoma and Texas from 2010 to 2014 to isolates from the plant.
PulseNet is focused on threats to human health, but WGS also is playing a central role in a nascent but even more ambitious project involving microbiome, the population of bacteria and other microbes found in all flora and fauna. Launched in January 2015 by Mars Inc. and IBM Research, the Consortium for Sequencing the Food Chain is billed as the largest metagenomics study of the food supply chain ever undertaken, leveraging artificial intelligence and big data to characterize raw materials, beginning with the factory and extending to the farm.
A year after its launch, a third organization joined the consortium: Bio-Rad Laboratories, a Hercules, Calif., provider of technologies to separate, purify and amplify biological materials. Bio-Rad contributes expertise in chromogenic and molecular testing to the investigation of genetic fingerprints.
Last June, Cornell University’s dairy research department joined the consortium. With its own dairy farm and processing plant, the school provides a closed loop supply chain for genetic sequencing of the biome of milk.
Cornell professor Martin Wiedmann provided a glimpse of what the combination of sequencing and bioinformatics might yield in a February food safety webinar sponsored by 3M. According to Wiedmann, spores that survive thermal pasteurization account for half of milk spoilages. By using WGS to sequence all the DNA in a generic milk sample and then compare that with prematurely spoiled milk, researchers could begin to apply metagenomics to build a predictive model to assess microbial shelf life and augment it with big data to identify variables in both the process and the field, such as weather conditions.
That would pave the way for various what-if questions, Wiedmann suggests, such as lowering storage temperatures a few degrees to extend shelf life or making adjustments in expiration dates based on changes in microbial populations as a function of seasonal weather changes.
If the definition of food safety can be stretched to include sour milk, then, yes, this constitutes better protection of the global food supply. It isn’t on a par with pinpointing the source of life-threatening bacteria, but it advances fundamental understanding of the nature of food, and that’s a good thing.