Applications of predictive microbiology to food packaging

INTRODUCTION

This section is to introduce the currently available predictive microbiology models, such as microbial growth, inactivation, survival, and others, to food applications including packaging. Because packaging may be the last step in food-processing operations, it may inherit the consequences of microbial transfer, growth, inactivation, and survival cascaded down during the food manufacturing. Packaging also functions as a protection barrier to environmental changes and potential abuses, which may cause food spoilage. In general, food-processing operations should have eliminated or reduced the harmful microbial counts to a safety level before the final packaging step, which provides a means to control the food qualities, including foodborne hazards. The microbiological shelf life can be estimated or predicted using mathematical models, if available and properly selected.

Microbial safety is one of the major quality control attributes for packaged food. Pathogen contamination in consumer food products happens every year. The development of effective methods to reduce and eliminate the potential microbial food hazards is the primary goal of food scientists, producers, and government at different levels. Several food pathogens have potentials to cause public foodborne hazards including Listeria monocytogenes, Escherichia coli O157:H7, Salmonella spp. and so on. The Centers for Disease Control and Prevention (CDC) estimated that about 2500 cases of listeriosis occurr each year, which results in 500 deaths in the United States. A survey in eight categories of ready-to-eat foods, collected over 14–23 months, from retail markets in Maryland and northern California FoodNet sites showed that 1.82% samples were positive with L. monocytogenes contamination (1). The E. coli O157:H7 outbreaks in spinaches and ground beef stress the needs of effective means to monitor and to ensure food safety, even more. The most recent Salmonella outbreaks in peanut butter indicated that microbial cross-contamination could occur during food production. Pathogenic public health hazard is always on the top of safety list for food producers and inspection agencies. Predictive microbiology is a useful tool to describe, predict, and assess the potential hazard in processed and packaged consumer food products.

The Food and Drug Administration (FDA) and the United States of Department of Agriculture (USDA) have established food-safety regulations that govern the manufacture and distribution of domestic and imported foods to ensure the safety of food supply chain. These regulations are often difficult to locate and/or complicated to comply even for food companies that have resources to employ food safety experts and to interpret the regulations. To promote food safety and reduce the financial burden, especially for small companies, the Predictive Microbiology and Bioinformatics for Food Safety and Security research group of USDA/Agricultural Research Service (ARS)/Eastern Regional Research Center (ERRC) has implemented a 3-year project to create the Predictive Microbiology Information Portal (PMIP) available on the Internet to help food producers and safety researchers better access food-safety information, regulations, and tools at no cost to the end users. The PMIP provides a wide range of food-safety-related information such as regulations, microbiological models, and microbiological data. The portal has been available to the public since September 2007, and it can be accessed at http://portal. arserrc.gov/. The portal was designed to make food regulations and tens of thousands of microbiological data accessible to the public.

The PMIP has been accessed million times from tens of thousands unique Internet Protocol (IP) addresses worldwide since its launch. The microbial modeling component in the portal, the USDA Pathogen Modeling Program (PMP), was developed and is maintained and regularly improved by the USDA/ARS/ERRC staff. The PMP is a user-friendly software that contain a set of mathematical models, which predict the behavior of major human pathogens in foods under selected environmental conditions commonly used in the food industry. The PMP is used throughout the world to assist food companies comply with food-safety regulations and to reduce the human illness risk through better food process and product designs. The PMP was downloaded about 5000 times annually and is routinely used by 30% of the food industry. The number of downloads of the PMP has increased 40% to an estimated 7000 times/yr since the PMIP launch, which indicates the portal is reaching out a wider customer base including the food packaging and shelf-life applications.

PREDICTIVE MICROBIOLOGY AND FOOD PACKAGING

The predictive microbiology (microbial predictive models) include growth, inactivation, surface transfer (or cross contamination), and survival models, which play important roles in the microbial food safety while tied in the food packaging design to reduce the microbial hazard. The transfer model may predict the pathogen transferred among process equipment or surfaces. The growth models show the potential growth of a specified pathogen under different conditions, e.g., temperature, pH, water activity, added preservative, and so on. The growth models may take into account of other environmental factors, which include modified atmosphere packaging (MAP) conditions, transportation, distribution, and consumer abuses if those factors are built in. Thermal or nonthermal process to reduce or eliminate microbial counts may be evaluated using the inactivation models. For the entire microbial safety assessment in a packaged product, a microbial transfer model can be applied to estimate the quantity of contamination, followed by the growth and/or inactivation models with designed process conditions to predict the potential pathogenic health hazards. With all information collected and available in the models, the users may select the parameters to match the packaging conditions and to predict the shelf life, for example, the fresh-cut packaged ready-to-eat vegetables. Other packaging-related models may apply to the model construction step, which describes the entire production processes to achieve the microbial shelf-life assessment.

Users also can utilize the models and database in the portal (PMIP) to reduce food-safety challenge studies for their new products. The PMP and Combase, which are accessible through PMIP, may provide the useful tools for microbial-safety-related shelf-life optimization and packaging design.

PREDICTIVE MICROBIOLOGY MODELS

For microbial growth, the sigmoid functions have been the most popular empirical models used to describe the microbial growth; one of such models is the modified Gompertz model, which is shown below (2):

where x(t) is the number of cells at time t; A is the asymptotic count as t decreases to zero; C is the difference in value of upper and lower asymptote; B is the relative growth rate at M; and M is the time at which the absolute growth rate is maximum or the inflexion point of the curve. Using the parameters in equation (1), the following terms can be derived to characterize the microbial growth:

The lag phase is the time for microbial to adjust to a new environment, followed by the exponential phase with a maximum growth rate until the available medium deprived or limited by other factors, then to the stationary phase. Baranyi and coworkers (3–5) introduced a mechanistic model, which includes the lag phase, the exponential growth phase, and the stationary phase. The explicit Baranyi model is expressed as the following:

where y(t) = ln x(t), yo = ln(xo), and v is the rate of increase of the limiting substrate, which is generally assumed equal to mmax. Parameter m is an index of the curvature before the stationary phase. The qo and xo represent the initial concentration of limiting substrate and cell number, respectively. mmax is the maximum growth rate. Baranyi models are more complicated than the modified Gompertz models and may be applied to the dynamic process conditions, e.g., temperature changes with time.

Other predictive models may include inactivation (thermal and nonthermal), cooling survival, and growth under influence of other factors, added or environmental. The growth models are classified as the primary model. The growth and/or inactivation models with parameters that interact with other factors are considered as the secondary models. Recently, the surface transfer models were developed to describe the cross contamination of food pathogens during slicing of foods, e.g., ready-to-eat deli meats, smoked salmon, and so on.

The transfer models typically only consider the pathogens transferred from one step to another step. The microbial counts may be performed at each individual step or at the end of one series of steps. Because it is a continuous and in a relative short time period, no growth or inactivation factors were built in this kind of modeling. The important factors will be the processing parameters used in the process flow, and therefore, empirical models were considered. Typically, a microbial count as a function of several operation parameters was presented. The surface transfer models shown below were the recently developed by Sheen (6) for L. monocytogenes transfer during slicing in two cross-contamination routes.

I: Slicing blade to meat product,

where Y is the log colony forming unit (CFU) per slice; X is the slice number index; and n is the initial microbial count in log CFU.

The general microbial inactivation model developed following the first-order kinetic chemical reaction described the microbial death reasonably well. However, the microbial inactivation could become complicated because of the microbial itself, environment conditions, and treatment applied to kill the micro-organism, especially for food pathogens. Some important parameters may be temperature dependent. The D and Z values, which are typically used in the thermal process to evaluate the thermal lethality and also can be adopted to other inactivation study. To simulate the inactivation, some studies used the curve-fitting method simply fit the experimental data. A nonlinear approach using the power law function that best represented the inactivation is shown as:

where p is the power. A concave or convex curve is represented by po1 or pW1, respectively.

If shoulder or tailing appears in the inactivation curve, then the model will become even more complicated. Researchers have demonstrated that the asymmetric Gompertz function or the mirror image of Baranyi growth model may fit certain nonlinear survival or inactivation curves well.

PMIP AND PMP SAMPLE CASE DEMONSTRATIONS

The homepage of PMIP shown below serves as the gateway to access other components available to the users. It is highly beneficial for the users to navigate the website and become familiar with all options and available data. Two examples are demonstrated.

EXAMPLE 1: THERMAL INACTIVATION OF PATHOGENS IN FOODS

The inactivation on microbial growth due to thermal (temperature) effect can be estimated by using the PMP models, which are available to certain food items. The user may select the heat inactivation (under the ‘‘ModelsBacterium’’ headline, online option), then the ‘‘food pathogen’’ to select the microbe, which leads the user to key in other parameters, like pH, salt content, sodium pyrophosphate, and targeted log reduction. The temperature also needs to be specified for the time-required calculation. The following figure shows the heat inactivation of 8 log L. monocytogenes reduction in ground beef requires 3.87 min (D-value = 0.48 min at 651C). If a lower temperature at 551C is selected with other factors remained the same, the process time becomes 159.93 min (D-value = 19.94 min at 551C). Therefore, to achieve the desired thermal inactivation, the PMP model may provide useful information to select the proper process with parameters fit to the product.

EXAMPLE 2: SHELF LIFE STUDY AND PREDICTION

The PMP models can be used to predict the shelf life of packaged foods. The user may find the parameters to match or closely fit the conditions of a food item and make the reasonable shelf-life prediction. For example, to predict the shelf life of vacuum-packaged seafood salad with the targeted pathogen like L. monocytogenes o100 CFU/g, the user may apply the following steps: (a) bacteria model; (b) L. monocytogenes; (c) growth anaerobic (shrimp and imitation crab salad); (d) select temperature, pH, time duration, time interval, initial level of L monocytogenes and level of concern; and (e) calculate growth data. Results will appear on the screen as the following, which indicates the shelf life is about 10 days. For comparison, one may increase the temperature to 81C and pH to 5.0, and the shelf life becomes 5 days.

The table also shows some useful information, us such as growth rate, generation time, lag phase duration, and so on. When the users apply a low level of L. monocytogenes (e.g., o0.1 log CFU), it is recommended that the worst-case situation used for food safety. This example is for demonstration only; the user should visit the regulatory component available on the same website and acquire the food regulations in different country and products. The L monocytogenes detection level in the United States is zero (negative) per 25 g and other countries may impose different tolerance levels.