<P>Objective 1. Demonstrate that the EZKnowzâ„¢ can detect Shiga toxin producing (stx) e. coli and listeria monocytogenes on spinach and netted cantaloupe.Feasibility question: Will we be able differentiate the odor signatures of these foodborne pathogens (stx E. coli and listeria monocytogens) from the food odors? We will measure the odor signatures from inoculated samples of spinach and netted cantaloupe against clean, non-inoculated controls and establish an odor signature indicating the presence of these pathogens on the selected foods.
<br>Objective 2. Evaluate the differences in odor signatures between the two pathogens and demonstrate that we can identify the one pathogen from the other based on their odor signatures on the selected food substrates.Feasibility question: Can we differentiate the signature for stx E. coli and listeria monocytogens from each other when inoculated and sampled on spinach and netted cantaloupe substrates?We will perform Partial Least Squares analysis of the odor signatures of the two pathogens selected and determine if they can be differentiated from each other. Additionally we will make an initial attempt to identify the biomarker VOC odor analytes that provide the differentiation.
<br>Objective 3. Demonstrate that we can identify the presence of at least one of the pathogens on at least one of the food lots using EZKnowzâ„¢ odor analysis at a level that is consistent with standards accepted by FDA using PCR or microbiological techniques.Feasibility question: Can the EZKnowzâ„¢ platform be used as a detector of foodborne pathogen in at least one specific case that has comparable sensitivity as FDA-approved tests at maximum acceptable levels.We will measure the odor signatures from inoculated samples and in parallel we will perform PCR and microbiological tests that are generally accepted protocols for measuring activity of these foodborne pathogens on these food types. In at least one case, we will demonstrate that EZKnowzâ„¢ can sense the presence of the pathogen at a level that is comparable to maximum acceptable FDA guidelines.Phase I MilestonesMS 1. Kickoff meeting with ANI, Nova Biologicals and Program Manager (conference call or web-meeting). Month 1MS 2. Bacteria species are obtained and ready for inoculation of food products Month 2MS 3. Food products are inoculated and VOC samples are acquired. Month 3MS 4. Data from odor samples are analyzed. Month 4MS 5. VOC odor results are correlated with microbiological results. Month 4MS 6. If needed, a second round of testing is completed. Month 6MS 7. Complete all analysis and achieve all Technical Objectives. </P>
<P>NON-TECHNICAL SUMMARY: In studies conducted for the Centers for Disease Control and Prevention, it was estimated that approximately 48 million new cases of food-related illness, resulting in 3,000 deaths and 128,000 hospitalizations, occur in the United States annually. These estimates, although lower than previous estimates, confirm that foodborne illness continues to be a problem. Together with the mortality and quality-of-life costs, foodborne illness places a tremendous burden on our society. It is even more of a burden in developing countries. There is no doubt that there is a need for low-cost, rapid detection of micro-organism in our post-harvest food processing and distribution network.Bacteria, as well as fungi produce VOC signatures that are a product of primary and secondary metabolism pathways, thus providing a means for quick, reagentless testing for foodborne pathogens. This proposal is responsive to the USDA SBIR Program Priorities as it addresses the need for a quick (minutes), point-of-test analysis for critical food borne pathogens.The Phase I efforts will provide the information needed to prove the feasibility of detecting foodborne pathogens on representative foods (spinach and netted cantaloupe) for two common foodborne pathogens (Shiga toxin producing E. coli and Listeria). We will measure the odor signatures from these inoculated samples against clean, non-inoculated controls. In parallel we will perform PCR and microbiological tests that are generally accepted protocols for measuring activity of these foodborne pathogens on these food types. The Phase I effort will demonstrate technical feasibility by establishing the level of detection of foodborne pathogens on food types that have a history of contamination with these pathogens in our food supply chain. The pathogens were also selected as there were among the most costly and common foodborne pathogens and are also of specific interest to USDA. The Phase II effort will expand our effort to include salmonella and campylobactor species. We will also develop an engineering prototype EZKnowzâ„¢ sensor platform that is optimized for food safety applications. One goal will be to achieve sampling and analysis time to less than 2 minutes. Should our program be successful, the technical benefit will be a low-cost, hand-portable, field-ready instrument that can provide fast (2 minute) and reliable (as compared to currently acceptable techniques) sensor for common foodborne pathogens. We will initially focus on bacterial pathogens but this could be expanded to viral pathogens as well. The economic benefit will be a tool that can be used in finding and managing foodborne pathogens at all levels of the food production chain, from field analysis to post-harvest distribution and sale. Even a decrease of 10% in the number of foodborne illness cases can lead to billion (USD) in savings an improved quality of life. It will even save lives. Benefits outside of food safety include recent studies which have shown the EZKnowzâ„¢ to be a potentially significant new technology for detecting multidrug-resistant pathogens in clinical as well as hospital-acquired infections, accurately identifying nosocomial pathogens. </P>
<P>APPROACH: Task 1: Design of Experiments (DOE)An initial set of experiments will be designed to achieve the technical objectives.Spinach samples as well as cantaloupe whole fruit (un-sliced) samples will be tested according to the study design for Challenge Testing of Refrigerated Ready-to-Eat Foods: http://www.hc-sc.gc.ca/fn-an/legislation/pol/listeria monocytogenes-test-eng.ph p.E. coli 0157:H7 ATCC 35150Listeria monocytogenes ATCC 7644Each sample will be tested after inoculation periodically (e.g. 4 hours) for up to 12-24 hours. It currently takes about 15 minutes to complete VOC collection and analysis for each test so we will be able to test spinach samples on one day and test cantaloupe samples on another. For each set of VOC tests, there will be a corresponding sampling for microbiological testing to compare odor results to CFU quantities to measure against FDA standards. The design of experiments also includes testing of food samples with no bacterial doping that will help us with baseline comparisons. Finally, bacteria will be grown on culture media for headspace VOC analysis to compare against bacteria grown on food samples.Task Responsibility: Dr. Fink (ANI) will be responsible for completing this task but he will work with Dr. Pavlovsky (ANI, data analysis) and Dr. Paul Pearce (Nova Biologicals, leader microbiological effort) to adjust to fit the constraints of resources and time.Task 2: Prepare for experimentsThis task will be focused on preparing for the bacteria-laced food experiments at Nova Biologicals.At ANI, this will be optimizing the odor collection approach and EZKnowzâ„¢ operating parameters to achieve the highest sensitivity of the tool with the least amount of saturation and/or interfering odors. Odors can be collected through a funnel that is placed over the sample.Disposable funnels will be used to avoid cross contamination. Some odor samples will be collected in GC-MS odor absorption tubes for later characterization by GC-MS (Cerium Lab, vendor).To demonstrate that odor markers from bacterial cultures will be separate from odors of spinach and netted cantaloupe, we compared the headspace odor spectra of E. coli collected at Nova Biologicals on an earlier project and compared them to odor spectra of spinach and cantaloupe.Nova Biologicals will prepare for plating the bacteria and make sure they have sufficient supplies to perform the microbiological tests. Nova will perform all Microbiology preparatory work including organism procurement, organism identification, inocula preparation and standardization. Nova will prepare all specimens for testing. Nova will inoculate all food samples. Nova will perform 96 total tests for detection and enumeration of E. coli and Listeria monocytogenes.Task Responsibility: Dr. Pavlovsky (ANI) will be responsible for completing this task. He will be assisted by Mr. Chip Thuesen. Dr. Paul Pearce will be responsible for the E. coli and Listeria monocytogenes inoculum preparation as well as performing the benchmark microbiological testing to FDA standards.Task 3: Algorithm development and data analysis.We plan to use algorithm approaches to perform two critical functions for this program: (1) to identify and limit the confounding effect of odors from the food sources; and (2) to down-select the most appropriate bacteria VOC markers that would identify the presence of bacteria on the food substrates. The objective is to find VOC markers that would be suitable across as many food groups as possible. The scope of Phase I is limited to spinach (representative of leafy greens) and netted cantaloupe (representative of melons). Phase II will expand our study to other food types and other pathogen species.Task 2 describes how we will collect data. The ANI team has already collected some bacterial VOC headspace odor signatures using the EZKnowzâ„¢ platform. Some VOC markers are already identified and consistent with previous studies using GC-MS instruments.Some VOC bacteria-related biomarkers are expected to be exclusively expressed by the bacteria (metabolic processes) and appear resistant to confounding factors (like, for example food odors). If we locate these unique chemicals in an unknown field sample, our software will make an easy but very reliable decision of the category (i.e., pathogen present) of the new sample.It is likely that some of the confounding VOCs will overlap with VOCs of interest The absolute abundance or magnitude of chemical interference may vary, and we will employ machine learning "pattern recognition" methods for the diagnosis of unknown samples compared to our library of these "masked" compounds. This will be performed by establishing a diagnosis model based on computational pattern models of the bacteria pathogen (established in this effort). Specifically we will establish diagnosis models that include:By concentrating information from the raw data into a number of principal components, PCR aims to overcome the major problems of ordinary least squares data fitting methods: (a) the sample number must be equal to or larger than the number of independent variables, (b) possible co-linearity in the independent variables could result in the ill-condition of a regression matrix and eventually lead to an unreliable solution, and (c) possible noise in the raw data would degrade the modeling effect. PLSR is another modified LSR, extracting uncorrelated latent variables from the original data (e.g. confounding effects like geography differences, or nutrient changes). The difference between PLSR and PCR is the extraction of latent variables in PLSR not only employs independent variables but also take into account dependent variables, so in general PLSR is expected to have an even better modeling effect. LDA is a linear classifier that expects to find the best separating line or plane between two groups of samples. LDA was designed for a two-class separation problem, but a stepwise linear discrimination functions can extend LDA to a multi-class separation problem.Task Responsibility: Dr. Pavlovsky (ANI) will be responsible for completing this task. He has significant experience with EZKnowzâ„¢ data analysis and library building (citrus disease states, pollen identification on Army STTR, bacterial VOC odor characterization).Task Exit: We will exit this task when a library is established for characterizing E. coli and listeria using the data on the first run.Task 4: Blind odor analysis studyAt the end of the program, ANI will return to Nova Biologicals for a blind study. Spinach and cantaloupe samples will be prepared (un-treated and laced with E. coli and listeria) by Nova Biologicals as in Task 1 but samples will be blind to ANI. Odor analysis will be performed by ANI using the EZKnowzâ„¢ instrument. Based on the library established in Task 3, we will characterize the data and make identification of which samples were contaminated with which bacteria and which samples were not contaminated. The goal will be to achieve detection levels that are equivalent to FDA standards used to reject food lots and signal contamination of the specific pathogen.Task Responsibility: Fink and Pavlovsky (ANI) will be responsible for completing this task. Nova Biologicals will prepare the samples and complete the confirmatory microbiological tests. </P>