Praedicere Possumus (PP), un applicazione per la microbiologia predittiva Prof. Mara Lucia Stecchini Department of Agricultural, Food, Environmental and Animal Sciences University of Udine mara.stecchini@uniud.it Iniziativa del Dipartimento di Biomedicina Comparata e Alimentazione dell Università degli Studi di Padova e del Corso di Studi in Sicurezza Igienicosanitaria degli Alimenti con sede a Vicenza 25 Maggio 2017 - Padova
This presentation will cover: Intervention strategies for food safety Food safety authority and food enterprise requirements An application to meet the demands of food safety authority and food enterprises: Praedicere Possumus (PP) Benefits to users: some examples Concluding thoughts
Intervention strategies for food safety Responsibility of society (through government regulation and food safety authority controls ) Responsibility of the food industry (through quality assurance/food safety interventions) Ensuring Food safety Who? Responsibility of the consumer
Ensuring Food safety How? Food legislation in the European Union and elsewhere includes both hazard - and risk - based approaches for ensuring safety. In hazard - based approaches, simply the presence of a potentially harmful agent at a detectable level in food is used as a basis for legislation and/or risk management action. Risk - based approaches allow consideration of exposure in assessing whether there may be unacceptable risks to health. If a hazardous organism is found it means something, but absence in a limited number of samples is no guarantee of safety of a whole production. Thus, product testing gives only very limited information on the safety status of a food. The value of risk - based approaches is becoming increasingly recognized
A risk - based approach and official controls Various risk ranking methods are available that prioritize risks Risk classification of food risk ranking or comparative risk assessment Risk ranking would be valuable on deciding to carry out risk based regulatory controls and in their prioritization
A risk-based approach and official controls Prevalence Concentration of the pathogens in food are important to determine risk for consumers (EFSA), but.growth of pathogenic microorganisms of public health importance could impact strongly on risk.
A risk-based approach and official controls FDA High risk foods designation must be based on a number of factors, including the likelihood that a particular food would support the growth of pathogenic microorganisms due to the nature of the food or the processes used to produce such food. EFSA The physico-chemical conditions that could permit pathogen growth and the treatments applied to the food products, are information used to rank a food into low risk, moderate risk, and qualified presumption of risk.
A risk-based approach and official controls Australia's Priority Classification System for Food Businesses" High risk foods are foods that "may contain pathogenic microorganisms and will normally support formation of toxins or growth of pathogenic microorganisms. Medium-risk foods are foods that "may contain pathogenic microorganisms but will not normally support their growth due to food characteristics; or food that is unlikely to contain pathogenic microorganisms due to food type or processing but may support formation of toxins or growth of pathogenic microorganisms. Low-risk foods are foods that "are unlikely to contain pathogenic microorganisms and will not normally support their growth due to food characteristics.
The growth potential can help to define a food as hazardous foods Laboratory (microbiological) challenge studies Predictive microbiology (modelling evidence) Systems used
A risk-based approach and official controls Growth or survival of bacteria in foods does not have the same impact on public health for all the hazards Some hazards must growth in the food before consumption to reach numbers sufficient for a significant probability of causing illness. Frontier Lifeline For other hazards, the numbers resulting from the initial contamination are usually sufficient to cause illness.
Intervention strategies for food safety Food enterprise demand for cost-effective and less time consuming ways to define food safety, and more specifically they are interested in: Understanding the factors that impact positively or negatively on the ability of pathogenic bacteria to survive or growth - PRODUCT Validating the control measures - PROCESS Verification of procedures (es. HACCP) - PROCEDURES Assessing the compliance of food to safety criteria- FINISHED PRODUCT TESTING Predicting an appropriate shelf life - STORAGE
Strategies for food safety A risk-based approach on food safety authority controls Production of safe food and the application of process control measures both will benefit by the application of predictive microbiology (modelling)
What is predictive microbiology (modelling)? The use of mathematics to describe the behavior of microorganisms
Predictive microbiology Progress based on two pillars Predictive microbiology Growth Curves USA 2008 Bioinformatics (database) Biomathematics (mathematical models)
What is predictive microbiology (modelling)? The basic idea underlying predictive microbiology is that the behavior of microorganisms is deterministic and able to be predicted from knowledge of the microorganism itself, and the immediate environment of the microorganisms. Nature methods, 7, 190-191, 2010
Predictive microbiology A microbiological predictive model can describe or predict the growth, survival or death of microorganisms in foods. These models typically relate the microbial growth, survival or death responses to the levels of the controlling factors, such as temperature, ph, water activity etc. FAO
Predictive microbiology The models are based on laboratory generated data. Ex: growth max = opt (T) (aw) (ph) (AH) Environmental factors Microbiological growth media (broths) are produced with different instrinsic parameters such as ph and salt level and are then inoculated with the relevant organism or cocktail of organisms. These broths are then stored at a range of temperatures and the microbial level present is assessed over time. These are known as kinetic growth models, as they allow assessment of the amount of growth that can occur.
Once the data is generated, statistical equations are fitted and these are then combined with a user-friendly interface.
USDA INTEGRATED PREDICTIVE MODELING PROGRAM (IPMP)
Predictive microbiology Modelling is only for specialist users? People in the official control and in food industry, even if they are unable to avail themselves of specific knowledge, are encouraged to use the modelling approach.
Could a modelling tool be made more accessible and affordable to benefit food enterprises? In this case.. The role of each factor that can contribute a hurdle effect should be taken into account and easily provided allowing users to define appropriate processing conditions and product formulations The outcomes of the models should be safety constraints, which could derive from government regulations (a growth boundary for a given pathogen) An answer to the question about the maximum safe shelf life should be provided based on the likely growth of the pathogen of concern The tool should meet the need for validation and verification of procedures (ex: HACCP) for controlling the hazard
Could a modelling tool be made more accessible and affordable to benefit food control authorities? In this case.. Relevant information about the potential risk associated with specific hazard/food combinations should be provided Information about current control measures, their effectiveness for controlling the risk should be made available The tool should be useful in assessing the compliance of food to safety criteria The tool should assist Auditors when they assess the safety of processes within a food business (i.e validation of procedures- ex. HACCP- for controlling the hazard)
Could a modelling tool be made more accessible and affordable to benefit users? In this case.. an application should provide an easy way to access modelling = functionality an application should enable users to retrieve information intuitively = instinctivity prediction should be simplified and intended for practical use = simplicity
Praedicere Possumus, PP This presentation describes a web-based application (Praedicere Possumus, PP), which is available on the web site of the University of Udine (Italy) http://praedicere.uniud.it/ This application is intended to evaluate the queried pathogen responses, which are translated into outcomes of practical use to meet the demand of producers and safety authorities. The modelling results could help users to improve the microbiological safety of food, to design new food, to explore microbiological risk in food.
Praedicere Possumus, PP http://praedicere.uniud.it/ For a practical use, the modular structure seems to us more appropriate The first module: generic module The second module: specific module Incresing complexity The third module: process module
Praedicere Possumus, PP This application, which provides a deterministic approach for prediction, contains a group of models that address: the growth-no growth the growth the thermal inactivation the non-thermal inactivation of 10 foodborne pathogenic bacteria B. cereus C. botulinum (prot.) C. botulinum (non prot.) C. perfringens E. coli L. monocytogenes Salmonella spp. Shigella spp. S. aureus Y. enterocolitica
PP: the models The framework of this application is the growth-no growth boundary model of Polese et al. (2011) which is based on the Gamma concept
Growth/no growth model for Growth-no growth boundary model Growth/no growth boundaries for Listeria monocytogenes in Pitina at 12 C with respect to a w and ph values (continuous line), 800 ppm CO 2 (dash-dot line), 195 mm lactate (dash-dot-dot line), phenols (grey area). Products to the left of growth boundary do not support the growth of the pathogen at the specified temperature
PP: the models C for T T, ph ph, a a, C MIC n i η (T T min ) (ph ph min ) (a w a w ) min 1 i=1 MIC i 0 1 P 1 1 otherwise P=0 min min w w min i i This growth-no growth boundary model has been introduced in the PP application for a number of reasons: to quantify the product of the cardinal optimal distances for growth probability, it uses a normalisation constant (η), which is species independent the model was constructed to be expanded to incorporate different preservative effects, allowing prediction whatever the complexity of the food system it is easy to apply
PP: the models This growth-no growth boundary model allows the distance between the growth boundary and specific product characteristics to be quantified by a P value (growth probability) When P is P>0.1, growth is considered likely and the pathogen population density is estimated with a kinetic approach, which uses a threephase linear model (Buchanan et al., 1997) When P 0.1, pathogens could die over time and non-thermal inactivation is simulated through a gamma-like model. Thermal inactivation is determined through log-linear curves (Bigelow and Esty, 1920)
PP: the options For developing and managing the factors that impact positively or negatively on the ability of pathogenic bacteria to survive or growth For predicting likely product safety through an appropriate shelf life The fractional contribution (f) of each inhibitory factor to P is estimated as a function of the difference between the actual level of the factor and the inhibiting value, adjusted for the sub-optimal interval of the factor The time dependent probability parameter (P t ) is described as a function of P and µ max and represents the change in growth probability over time n (T - T min ) (ph - ph min ) (a w - a w min ) C i Π = ρ 1 - (Topt - T min ) (phopt - ph min ) (a w opt - a w min ) i=1 MICi
PP: the internal database The internal database of PP application includes growth parameters derived from Combase Predictor and thermal inactivation parameters from van Asselt and Zwietering (2005). Non-thermal inactivation parameters were obtained by using a newly developed conservative model (paper in preparation).
PP: The modules The menu structure is via modules, which assist the operator to organize information around the user s specific needs
PP: The modules Generic module The first module provides the opportunity to predict the growth probability, the growth and the thermal/non thermal inactivation of all the 10 pathogenic bacteria under different conditions (T C, ph, a w ), including a gamma factor defined by the user. This module helps the operator to identify the hazards able to grow and the pathogen levels in the selected conditions, which can include processing and/or storage situations.
PP: The modules Generic module Useful option: the time-dependent growth probability (P t ) The variable outcome P t, which takes into account the storage time in a time dependent probability parameter, allows the user to evaluate the probability of growth related to a specific time. This option is expected to assist users in the estimation/validation of the shelf-life of food under safety constraints.
PP: the time-dependent growth probability (P t )
PP: The modules
Storage time of 5 days at 8 C Storage time of 3 weeks at 8 C
What is the impact of puree main characteristics, thermal treatment and storage temperatures (5 C, 8 C) on the growth of pathogens? puree (ph 6.0; a w 0.99; thermal treatment at 72 C for 4 min; storage for 240 h) Storage at 8 C for 240 h (10 days)
What is the impact of puree main characteristics, thermal treatment and storage temperatures (5 C, 8 C) on the growth of pathogens? puree (ph 6.0; a w 0.99; thermal treatment at 72 C for 4 min) Storage at 5 C for 240 h (10 days) Storage at 5 C for 120 h (5 days)
What is the impact of gnocchi main characteristics, thermal treatment and storage temperature/time on the growth of pathogens (60 days at 4 C/7 days at 10 C)? gnocchi (ph 5.1, a w 0.975, thermal treatment at 80 C for 1 min) Storage at 4 C for 1440 h (60 days) 4 C 4 C
What is the impact of gnocchi main characteristics, thermal treatment and storage temperature/time on the growth of pathogens (60 days at 4 C/7 days at 10 C)? gnocchi (ph 5.1, a w 0.975, thermal treatment at 80 C for 1 min) Storage at 10 C for 168 h (7 days) 10 C
Challenge versus modelling Gnocchi Storage at 8 C for 816 h (34 days) Time (days) at 8 C Observed Bacillus cereus response % positive Log CFU/g (Log CFU/g increase >0.5) 0-2.81±0.08 17 25 2.99±0.56 34 100 5.21±0.52
How to select an appropriate shelf life? Since different storage times could influence the level of the pathogens, the duration of shelf-life should be determined based on the extent to which the pathogen of concern is able to grow. Using the predictive element P t with 0.1 cut-off, it may be possible to establish a shelf-life duration time that will prevent the growth of the pathogens.
What is the impact of ph on the shelf life of pesto? ph 5.3, aw 0.96; TT 90 C for 1 min; 3 weeks of storage at 8 C ph 5.3
What is the impact of ph on the shelf life of pesto? ph 5.1, a w 0.96; TT 90 C for 1 min; 3 weeks of storage at 8 C ph 5.1
PP: The modules Specific module In a second module, for each specific hazard, additional environmental factors are taken into account for predicting probability of growth / growth / inactivation, including also organic acids, food additives and microbial interaction. The introduction of the (f) option, which quantifies the fractional contribution of each inhibitory factor to growth probability, can assist the user in defining processing and storage conditions to attain a desirable food safety level. Useful option: the fractional contribution (f) of each inhibitory factor to P
PP: The modules
Listeria monocytogenes
Temp ph aw lactic acid Temp ph aw smoke
PP: The modules Process module A further module enables the user to describe specific food production processes, including different steps/time, which are modelled such as P, P t, growth and inactivation. For each processing step/time, quantification of the contribution of each inhibitory factor on the pathogen behaviour helps the user to optimise a process and to identify the critical control points.
PP: The modules
Listeria monocytogenes
Listeria monocytogenes
Are specific artisanal processing conditions favorable or not favorable for a pathogen growth? How to determine the compliance of a RTE food with the EU Regulation criteria taking into consideration the effect of processing? PITINA Pitina is a low-acid non-fermented meat product made by small producers who follow a traditional local recipe from North-East Italy. It is actually consumed raw after being processed for about 30 days. Pitina is an RTE which is not subjected to a significant pathogen reduction during processing.
For its final charachteristics Pitina is unable to support the Listeria growth At the end of processing Pitina was found to be unable to support the growth of L. monocytogenes at refrigeration temperature (4 C), at the most likely storage temperature (6 C) and at abused temperature ph a w Phenols (ppm) Pitina final characteristics CO 2 (ppm) NaNO 3 (ppm) Lactic acid (mm) 5.86±0.11 0.913±0.009 57 800 30±9.0 195±15 (12 C), with ph and a w being the substantial factors contributing to stability. www.italyeatfood.it
But..what is the impact of processing on the pathogen growth? Comparison of observed and predicted responses of L. monocytogenes in Pitina samples Pitina processing steps Observed L. monocytogenes growth response Predicted L. monocytogenes growth response Observed Log CFU/g P c P d t % positive a increse b predicted log cfu/g increse Drying (I) 0 0.16 0.25 0.04 0.16 ( 0.03 e ) Smoking (I) 0-0.00 0.04 - Smoking (II) 0-0.00 0.04 - Drying (II) 0-0.00 0.04 - Shelf life 0-0.00 0.04 - The safety of Pitina was almost unaffected by the processing procedures a percentage (%) of Listeria positive samples (Log increase >0.5 Log CFU/g) b log CFU/g increase was calculated as the difference between the log concentration (mean value) reached in each stage and the initial inoculum level c Response of the PP application in term of probability of growth, P 0.1 unlikely growth conditions d Response of the PP application in term of probability of growth related to a specific time, P t 0.1 unlikely growth conditions e standard error
Are specific artisanal processing conditions favorable or not favorable for a pathogen growth? SALAME without STARTER
Predicted growth and growth probability of L. monocytogenes in salami samples and fractional contribution (f) of each of the inhibitory factors taken into account. Predicted growth and growth probability of L. monocytogenes obtained by simulating potential scenarios which include the contribution of starter cultures.
Are specific artisanal processing conditions favorable or not favorable for a pathogen growth? Applying PP to potential scenarios that included the contribution of starter cultures, the onset of Listeria inhibition was predicted earlier, preventing the pathogen to exceed its critical limit. SALAME without STARTER
Fig. 1. Graphic representation of how PP can be applied for process and product safety evaluation. a: safe conditions; b: unsafe conditions. Model interpretation criteria: P: probability of growth, when P is 0.1 the product is considered to be unable to support growth, whereas when P is > 0.1 the product is considered able to support growth. Pt: time-dependent probability of growth, when P is > 0.1 and Pti is > 0.1, growth (>0.5 log CFU/g) is possible up to time i. P: dash-dot lines; Pt: dashed lines. Predicted population size (log N) (log CFU/g): solid lines. Safe areas: shaded zones on the left (related to Pt) and shaded zones on the right (related to P).
PP would be used to evaluate: Growth chance of the pathogen during food processing Intervention on processing Reformulation of the product
PP would be used to evaluate: Growth chance of the pathogen in the final product Classification of the product (high risk, moderate risk, low risk) Shelf life
Conclusions In conclusion PP, which could be useful for: evaluating the growth chance of a pathogen, quantifying the combined effect of various hurdles on the probability of growth, defining combination at which growth ceases, selecting shelf life is provided to set up: risk-based official controls food industry control measures Such application may lead to a more realistic estimation of food safety risks, and can provide useful quantitative data for the development of processes that allow production of safer food products (EFSA, 2012).
Concluding thoughts While we acknowledge the high variability of the factors affecting the pathogen behaviour in food and the effectiveness of the stochastic approach to deal with this variability we believe it could be more practical and realistic in those circumstances to apply a simplified approach, such as PP, which does not require modelling specialists Calatrava bridge in Venice We hope that this application will represent a bridge between producers and food safety authorities and will assist in their common commitment towards the production of safer food
Thank you for your attention Acknowledgments Pierluigi Polese Manuela Del Torre Praedicere Possumus (PP) is a free web-based software tool (http://praedicere.uniud.it/) for predictive microbiology, which has been developed within the MIUR Project Proof of Concept Network (PoCN), managed by AREA Science Park (Trieste, Italy). Mara Lucia Stecchini Department of Agricultural, Food, Environmental and Animal Sciences, University of Udine, Udine, Italy mara.stecchini@uniud.it