An official website of the United States government.

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Identification of Persistent Foodborne Bacterial Pathogens from Environmental Sampling Results with Statistical and Whole Genome Analyses

Objective

<p>The goal of my work is to identify when foodborne pathogens persist, by differentiating when pathogens repeatedly present in food production, processing, and handling environments are more likely from point sources or random sources of pathogen contamination. I will use three complimentary approaches: </p>
<p>(i) use expert opinion to classify sampling results as demonstrating persistence, </p>
<p>(ii) statistically infer persistence from environmental sampling data, and </p>
<p>(iii) apply whole genome sequence epidemiology to identify persistent bacterial pathogens from phylogenetic results. </p>
<p>Each objective uses different aspects of persistence biology, observational capacity, and observation interpretation to reach their conclusion. Each method will be evaluated against subsequent methods to develop scientifically valid methods with variable sophistication to identify persistent foodborne pathogens in food associated environments. The work will </p>
<p>(a) apply methods of clinical microbiology and epidemiology to push the limits of the analysis of existing environmental sampling data, </p>
<p>(b) justify when operations must increase sampling to inform environmental control strategies for foodborne pathogens, and </p>
<p>(c) dramatically expand genomic data on retail associated L. monocytogenes.</p>

More information

<p>NON-TECHNICAL SUMMARY:<br/> Listeria monocytogenes is an opportunistic foodborne pathogen responsible for approximately 250 deaths per year in the U.S., making it the second greatest cause of deaths due to known bacterial pathogens transmitted through food in the U.S. Annual costs of morbidity and mortality from listeriosis are estimated at $2.6 billion, ranking this pathogen behind Salmonella and the parasite Toxoplasma. Nearly 90% of listeriosis cases have been attributed to consumption of contaminated read-to-eat deli meats, most from food sliced at retail, providing a focal point of listeriosis control. Reviews have demonstrated that L. monocytogenes can persist in food associated environments for long periods of time, and that these persistent strains have been linked to outbreaks of foodborne disease. Industry has responded by promoting the Seek and Destroy method
where increasing aggressive sampling and cleaning follow recurrent positive samples to find the internal point-source of persistence. Niches are eliminated through improved sanitation, equipment, and process design. Strains may also persist due to external point-sources of repeated reintroduction, such as when 4 months persistence of a particular L. monocytogenes strain on raw fish in a processing plant was eliminated by changing from on particular fish supplier. Bacterial subtyping methods distinguish unrelated environmental isolates of L. monocytogenes, but researchers are left with applying non-standard, post-hoc rules to flag persistent strains. The lack of a systematic method to identify persistent strains is the knowledge gap that I will address with this project. To study L. monocytogenes persistence in food associated environments I will use the retail deli as a model system for
differentiating point-sources of persistent subtype contamination from when random reintroduction causes repeated observation of the same subtype. The goal of my work is to differentiate when foodborne pathogens repeatedly present in food production, processing, and handling environments are most likely from point sources or random sources of pathogen contamination. I will use three complimentary approaches: (i) use expert opinion to classify sampling results as demonstrating persistence, (ii) statistically infer persistence from environmental sampling data, and (iii) apply whole genome sequence epidemiology to identify persistent bacterial pathogens from phylogenetic results. Each objective uses different aspects of persistence biology, observational capacity, and observation interpretation to reach their conclusion. Each method will be evaluated against subsequent methods to develop
scientifically valid methods with variable sophistication to identify persistent foodborne pathogens in food associated environments. Activities in this proposal will enhance control of foodborne pathogens in food associated environments. The work will (a) apply methods of clinical microbiology and epidemiology to push the limits of the analysis of existing environmental sampling data, (b) justify when operations must increase sampling to inform environmental control strategies for foodborne pathogens, and (c) dramatically expand genomic data on retail associated L. monocytogenes. Results are immediately relevant to retail food systems, which have unique challenges in implementing the industry standard Hazard Analysis Critical Control Point (HACCP) system for proactive food safety, by improving their use of environmental sampling to monitor and eliminate foodborne pathogens from the
environment.
<p>APPROACH:<br/> Expert Opinion: Studies of L. monocytogenes persistence generally use simple heuristics that vary between authors leaving food safety professionals without systematic methods to identify and control persistence. Expert elicitation can fill this gap using systematic processes to quantify expert judgment related to the probability and uncertainty of phenomena. This method pursues the hypothesis that food safety experts have systematic methods to identify bacterial persistence, making their classification a valid method for rapid identification of bacterial persistence from environmental sampling data. Machine learning techniques will be used to extract rules for persistence classification because they systematize multi-dimensional input and response data. The proposed workflow will involve (a) feature extraction, (b) dimensionality reduction, and (c)
classification modeling. Methods will be optimized for confidence in the final classification, simplicity of interpretation, and efficient prediction. This objective will build a store-level model to classify L. monocytogenes subtypes from environmental monitoring of retail deli operations as persistent or not based on a synthesis of expert opinion. Data will inform a preliminary spatial model for identifying bacterial persistence. Stratification of experts into different sectors will suggest if academic, industrial, and regulatory experts use qualitatively different methods to identify persistence. Statistical inference: Current environmental sampling analyses cannot systematically differentiate point-sources of persistent contamination from random reintroduction of pathogens in food associated environments. Both discrete statistical distribution tests and hidden Markov models (HMMs)
will be used to infer bacterial persistence. Test statistics will be evaluated using simulated environmental sampling data ranging across (a) ecological parameters, (e.g. store prevalence, variable subtype distributions) (b) monitoring parameters, (e.g. test frequency, test specificity) and (c) different signal of persistence. Statistics will be evaluated by Receiver Operating Characteristic curves and sensitivity analysis to determine their range of applicability. This objective will produce well-characterized statistical tests that food safety processional can use to infer persistence based on their environmental sampling data. Simulation studies inform environmental monitoring program design based on test sensitivity and specificity. Whole genome sequencing (WGS): Here WGS will be applied to analysis of environmental sampling data using the conceptual model that persistent subtypes
will show more phylogenetic relatedness than random members of the subtype. The full genome of bacterial isolates that appear to be persistent or sporadic based on earlier methods will sequence using modern high-throughput techniques, a phylogenetic tree constructed to determine the population structure of those isolates, and persistent isolates will be identified as those isolates showing statistically more relatedness than background members of the same bacterial subtype. This objective will provide a both a gold standard method to identify bacterial persistence from environmental sampling data and validate simpler methods of expert-judgment and statistical inference for routine industry use. WGS data will be compared to the 100K Genome Project's growing database of foodborne pathogen genomes (http://100kgenome.vetmed.ucdavis-.edu/about/index.cfm) to investigate the evolutionary
and metabolic determinants of persistence.

Investigators
Stasiewicz, Matthew
Institution
Cornell University
Start date
2013
End date
2015
Project number
NYC-143574
Accession number
1000542
Commodities