Location: Microbial and Chemical Food Safety
2020 Annual Report
Objectives
Mathematical models that predict behavior of pathogens in food can be used to verify critical control points in Hazard Analysis and Critical Control Point (HACCP) programs. For example, they can be used to assess whether or not a process deviation results in a one log cycle increase of Clostridium perfringens during cooling of a cooked meat product during commercial processing. Models that predict behavior of pathogens can be integrated with data for pathogen contamination to predict dynamic changes in pathogen prevalence and number in food across unit operations of a production chain. Predictions of consumer exposure can then be used in a dose-response model to form a process risk model that predicts consumer exposure and response to pathogens in food produced by specific scenarios. Process risk models have great potential to improve food safety and public health by providing a better assessment of food safety and identification of risk factors. In the past, we have developed predictive models and process risk models that have proven highly useful in providing regulatory agencies and the food industry with an objective means of assessing food safety and identifying risk factors. The goal of the proposed research is to elevate that successful effort to the next level of sophistication by considering additional variables and developing new and improved models and more effectively transferring this new research to the food industry by providing updated and improved versions of our software products: the Predictive Microbiology Information Portal, ComBase, and the Pathogen Modeling Program.
1: Develop and validate predictive models for behavior of stressed and unstressed pathogens in food with added antimicrobials. This includes development of validated dynamic models for spores and vegetative foodborne pathogens for evaluating heating and cooling process deviations.
2: Develop and validate process risk models for higher risk pathogen and food combinations.
3: Expand and maintain the ARS-Pathogen Modeling Program and Predictive Microbiology Information Portal. Continue to support the development and utilization of ComBase with our associated partners the Institute of Food Research (IFR) and the University of Tasmania (UTas) as an international data resource.
Approach
Effects and interactions of time, temperature, pH, acidulant, water activity, humectant, or preservatives (phosphates, organic acid salts, and nitrite) in meat and poultry products, as well as in rice, beans, and pasta will be assessed to collect kinetic data for pathogens (Listeria monocytogenes, Escherichia coli O157:H7, Staphylococcus aureus, Salmonella, Clostridium perfringens and Bacillus cereus). Kinetic data will be modeled using primary and secondary models. Predictive models performance will be evaluated using the acceptable prediction zone method. Once validated and published, predictive models will be incorporated into the Pathogen Modeling Program and data will be archived in ComBase.
Kinetic data for development of predictive microbiology models for survival and growth of pathogens (Salmonella, E. coli O157:H7, Campylobacter jejuni, and Listeria monocytogenes) on higher risk food (tomatoes, lettuce, raw milk, and crab meat) will be obtained in inoculated pack studies. Pathogens will be enumerated on higher risk food during storage trials using an automated miniature most probable number method. Kinetic data will be modeled using neural network modeling methods and models will be validated against independent data using the acceptable prediction zone method. Whole sample enrichment real time polymerase chain reaction (WSE-qPCR) will be used to obtain data for prevalence, number, and types of pathogens on higher risk food. Predictive microbiology models and contamination data obtained by WSE-qPCR will be integrated to form process risk models that predict consumer exposure and response to pathogens on higher risk food produced by different scenarios.
All new models will be added to both versions of the Pathogen Modeling Program. A link to ARS, Poultry Food Assess Risk Models website will be provided in the portal. Combase will be made compatible with the PMP.
Progress Report
Progress was made on all objectives, all of which fall under National Program 108 – Food Safety, Component I, Foodborne Contaminants. Progress on this project focuses on Problem F, Predictive Microbiology/Modeling: Data Acquisition and Storage.
Under Objective 1, experiments were conducted to assess the ability of Bacillus cereus spores to germinate and grow at isothermal temperatures from 10 to 49°C in rice/chicken (4:1), rice/chicken/vegetables (3:1:1), rice/beef (4:1), and rice/beef/vegetables (3:1:1). Once completed, predictive models for growth of B. cereus at temperatures applicable to cooling of cooked products will be developed. The growth data/predictive models on the safe cooling rate of foods will provide the food industry means to assure that cooked products are safe for human consumption.
Under Objective 1, experiments were conducted to determine Staphylococcus aureus growth at various isothermal temperatures from 10 to 54°C. Predictive model for growth of S. aureus at temperatures applicable to low temperature long time cooking of food products will be developed. The growth data/predictive model on the safe cooking rate of foods will provide the food industry means to ensure safety of cooked products.
Under Objective 2, data collection was completed for development and validation of a predictive model for growth of Salmonella on chicken liver as a function of times and temperatures observed during meal preparation. In addition, data collection was initiated for development and validation of a predictive model for growth of Salmonella on chicken gizzard as a function of times and temperatures seen during meal preparation. These models will fill important data gaps in process risk models (PRM) for Salmonella and chicken by-products. The PRM can be used in quantitative microbial risk assessments that are used as the scientific basis for new food safety policies aimed at safeguarding public health.
Under Objective 2, data collection was initiated for mapping Salmonella contamination (prevalence, number, serotype) on the chicken carcass. These data will be used to identify hot spots of contamination that will help the chicken industry better target interventions (antimicrobial rinses) to specific areas of the chicken carcass to more effectively reduce or eliminate Salmonella contamination. The result will be less risk of foodborne illness, less chance of product recalls, and better public health.
Under Objective 2, a new software tool called vault (ValT) was developed and published. The goal of ValT is to help scientists properly apply the test data, model performance, and model validation criteria of the acceptable prediction zones method. Proper validation of predictive models for foodborne pathogens is important because it increases confidence of end users in the food industry and regulatory agencies for using models to make important food safety decisions. In addition, it improves accuracy of model predictions by helping scientists identify and repair prediction problems in models before they are published and distributed to end users in the food industry and regulatory agencies.
Under Objective 2, a process risk model for Salmonella and ground chicken was developed and published. Two determinations were made from this study: (1) the method (rinse aliquot) used to detect Salmonella in poultry meat underestimates prevalence and thus, provides an inadequate prediction of food safety; and (2) there are also other factors (pathogen number and type (virulence), temperature abuse, undercooking, cross-contamination, host resistance) besides Salmonella prevalence that determine risk of foodborne illness (salmonellosis) outbreaks. Just like assessing risk of riding a car based on one factor (presence of seat belts) is not a good idea, assessing risk of foodborne illness based on one factor (pathogen prevalence) is not a good idea. Thus, a new holistic approach to food safety is needed; one that uses multiple risk factors to identify unsafe food before it is shipped to consumers.
Under Objective 2, a commercial software tool (NeuralTools) that is an add-in program for a common spreadsheet program (Excel) is was used to develop an Artificial Neural Network (ANN) for predicting growth of Salmonella in laboratory broth as a function of time, temperature, previous pH, and pH. Results of this study are important because they showed that ANN can be used to efficiently learn patterns in a large and complex dataset (1,513 data points) to accurately predict a complex three-phase, sigmoid-shaped growth curve across multiple combinations of interacting variables important for growth of Salmonella in food. Thus, ANN has a bright future as a new tool in the food safety arsenal aimed at protecting public health from pathogens that contaminate and grow in our food.
Under Objective 3, ComBase is an international microbial modeling database. It continues to grow in size, relevance and impact for the food industry, government and international researchers who seek to improve global food safety and collaborations. In the past 12 months, there were 66,401 user sessions, 73,160 registered users (current average of 11,020 new registered users per year), 1,094 new data records, and the top 10 countries using ComBase were Spain (18.38%), Italy (8.59%), United States (7.77%), United Kingdom (6.06%), Canada (4.71%), Colombia (4,39%), Mexico (4.29%), Netherlands (4.07%), Japan (3.51%), and France (2.99%).
Accomplishments
1. Safe salad. Salad is a popular side dish served with chicken. However, cross-contamination of salad with Salmonella from utensils (cutting board, knife, hands) used to process raw chicken for cooking followed by growth of the pathogen on salad before serving could lead to foodborne illness. Growth of a low number (7 cells) of a chicken isolate of Salmonella Newport on Romaine lettuce as a function of times (0 to 8 h) and temperatures (16 to 40°C) seen during meal preparation and serving was investigated and modeled by ARS researchers in Princess Anne, Maryland. A computer model developed from the study can be used to predict growth of Salmonella on salad under different scenarios of meal preparation and serving before consumption. This new knowledge fills an important data gap in risk assessments conducted by regulatory agencies who protect the food supply and public health.
2. Proper means for cooling of cooked foods. Inadequate rate and extent of cooling is a major food safety problem. Scientists at Wyndmoor, Pennsylvania, assessed the ability of Clostridium perfringens and Clostridium botulinum spores to germinate and grow in cooked pork and beef, respectively, at temperatures applicable to cooling of cooked products. The growth data/predictive models developed on the safe cooling rate will provide the food industry means to assure that cooked products remain pathogen-free and are safe for human consumption.
3. Modeling heat resistance of Listeria monocytogenes in beef. Adequate heat treatment destroys L. monocytogenes and is the most effective means to guard against the potential hazards in sous vide cooked ground beef. Due to public health concerns regarding toxicity of synthetic chemicals and microbial resistance to such preservatives, consumers these days are increasingly demanding natural products. ARS researchers at Wyndmoor, Pennsylvania, assessed the efficacy of lauric arginate on the reduced heat resistance of L. monocytogenes in sous vide cooked ground beef. The predictive model developed will assist food processors to design appropriate thermal processes for the production of safe sous vide beef product with extended shelf life.
4. ComBase, an international microbial modeling database. A new search feature has been added to the Browser; each record now indicates the date that the record was added to ComBase; an improved and simpler data donation template, plus instructional videos, have been added to the Data Submission page; enhanced messaging on website to promote data donations; each data record now indicates the number of times it has been viewed and downloaded; a YouTube channel and tutorials are now available; a private data section with ComBase is available to embargo data until a publication has been released; the ComBase Predictor was changed to ‘Broth Models’ in the menu, so that it better aligns with the separate suite of ‘Food Models’; a new feature allowing CB data to be added (over-laid) on ComBase Predictor graphs; updates to Perfringens Predictor and inactivation models per USDA-FSIS requests (hyperlinks to FSIS documents, increasing time-temp input capacity to ~500 data points, specific directions about how to measure core temperature, allowing a maximum 10 degree F jump in cooling temp for Perfringens Predictor); displayed all three kinetic parameters—lag, growth rate, MPD--for ComBase Predictor growth model outputs; added a reset button for model default lag time; integrated API feature to link model predictions to 3rd party software; Social media accounts are on Facebook (163 followers), LinkedIn (4,175 connections), and Twitter (1,655 followers). ComBase assists users in predicting and improving the microbiological safety of foods as well as in assessing microbiological risk in foods. Thus, ComBase saves the food industry millions of dollars a year by reducing the need for costly microbiological tests as well as helping to prevent recalls and foodborne illness.
Review Publications
Cosansu, S., Juneja, V.K., Osoria, M., Mukhopadhyay, S. 2019. Effect of grape seed extract on heat resistance of Clostridium perfringens vegetative cells in sous vide processed ground beef. International Journal of Food Science and Technology. 120:33-37. https://doi.org/10.1016/j.foodres.2019.02.014.
Leng, J., Mukhopadhyay, S., Sokorai, K., Ukuku, D.O., Fan, X., Olanya, O.M., Juneja, V.K. 2019. Inactivation of Salmonella in cherry tomato stems cars and quality preservation by pulsed light treatment and antimicrobial wash. Food Control. 110:107005. https://doi.org/10.1016/j.foodcont.2019.107005.
Zhou, S., Jin, Z.T., Sheen, S., Zhao, G., Liu, L.S., Juneja, V.K., Yam, K. 2020. Development of sodium chlorite and glucono delta-lactone incorporated PLA film for microbial inactivaton on fresh tomato. Food Research International. 132:1-7. https://doi.org/10.1016/j.foodres.2020.109067.
