1. 7 th Dubai International Food Safety Conference & IAFP’s 1 st Middle East Symposium on Food Safety EXAMPLES OF EXISTING MODELLING TOOLS FOR TRACKING MICROBIAL HAZARDS IN FOOD CHAIN Moez SANAA & Ewen TODD

2. QRAMODELS“PRODUCTION-TO-CONSUMPTION” Cross-contamination and Recontamination models Dynamic models for predictive microbiology including Growth & Survival Specific to the food matrix Consumption patterns Panels, Health status Starting material Management of the primary production Pre-harvest activities Primary production models Quantitative analysis of raw material quality data / farm release models statistical analysis procedures RISK Raw Materials TransportRetailConsumerTransport Process/ Food packaging Thermal transfer models

3. MODEL GENERAL PRINCIPLESEXAMPLE M ILK PRODUCTION 1.Raw Milk contamination Growth during transport and storage During processing: reduction/survival/Growth 2.Contamination during processing Recontamination Transfer of organisms from plant environment to cheeses Cross contamination transfer of microorganisms from one cheese to another caused by direct or indirect contact Bacteria fate 1.Products/Environment 2.Growth/stress 3.Detection / Response: detect and respond to “incidents”

4. COMPARTMENTAL MODEL Cheese processing room Passageway Smearing machine room Packaging room Presence of bacteria colonies in different compartments : milk (cells/Liter), Products (colonies/Product), Environment (colonies), Machines (colonies) State of compartment C at time t: C t = (a i, b i ), i = 1 to n a i = size of the colony i (cells) b i = Latency specific to colony i n = number of colonies

5. STEP S+1 COMPARTMENTAL MODEL STEP S Lot K1 Cheese Machine Environment S Lot K2 Cheese Environment S+1 Transfer of colonies Intra-lot and inter-lot contaminations Intra-step and Inter-steps contaminations

6. MODEL THE TRANSFER OF COLONIES Machine Environment Cheese p me p cm p mc

7. 7 th Dubai International Food Safety Conference & IAFP’s 1 st Middle East Symposium on Food Safety Moez SANAA FATE OF THE MICRO-ORGANISM IN FOODSTUFFS (PREDICTIVE MICROBIOLOGY MODELS) IMPACT OF FOOD TECHNOLOGY

8. O UTLINE Primary growth models Classical models Microbial interactions Secondary growth models Cardinal models Growth/no Growth boundary Lag times models Model validation

9. G ROWTH PHASES E NVIRONMENT CONDITIONS ARE CONSIDERED CONSTANT Time ln(x) Lag (latency) exponantial Stationary Death ln x max ln x 0 

10. 0100200300400 5 6 7 8 9 10 0100200300400 5 6 7 8 9 10 temps (h) log 10 ufc.ml -1 0100200300400 5 6 7 8 9 10 0100200300400 5 6 7 8 9 10 0100200300400 5 6 7 8 9 10 exponential Gompertz logistic Baranyi Rosso log 10 x 0 = 5.90 lag = 39.9 h  max = 0.037 h -1 log 10 x 0 = 5.86 log 10 x max = 9.54 lag = 50.3 h  max = 0.043 h -1 log 10 x 0 = 5.60 log 10 x max = 9.42 lag = 38.1 h  max = 0.042 h -1 log 10 x 0 = 5.85 log 10 x max = 9.32 lag = 47.5 h  max = 0.040 h -1 log 10 x 0 = 5.90 log 10 x max = 9.35 lag = 39.7 h  max = 0.037 h -1

11. F ACTORS THAT AFFECT GROWTH Biotic environment Competition for nutrients, production of specific inhibitors (bacteriocins), alteration of the environment Abiotic environment Temperature, oxygen levels, specific preservatives (e.g. nitrite, organic acids, smoke components), space limitation, diffusion of nutriments, etc. Strain differences

12. M ICROBIAL INTERACTION Giménex & Dalgaard 2004, Mejlholm & Dalgaard 2007

13. C OMPARING TWO CONSTANT ENVIRONMENT CONDITIONS Time (h) ln x ln x max lag 1 ln x 0   lag 2 pH 1 = pH 2 aW 1 = aW 2 T opt = 37°C, T min = 2°C T 1 = 25°C, T 2 = 15°C x max = x max1 = x max2 x 0 = x 01 = x 02

14. S ECONDARY G ROWTH MODELS Environmental Factors "CARDINAL Model"  max (h -1 ) 01020304050 0 0.2 0.4 0.6 0.8 1 1.21.4 température (°C) pH = 7 pH = 6 pH= 5.5 pH = 5 46810 pH T = 37°C T = 25°C T = 15°C T = 10°C

15. S ECONDARY G ROWTH MODELS Cardinal temperature model

16. F ULL CARDINAL MODEL

17. C ARDINAL MODEL ASSUMPTIONS Optimal growth is a characteristic of microbial strain specific to food matrix Cardinal parameters are strain specific Could be assessed using broth media Strain variability could be captured by varying cardinal parameters

18. pHT°C Aw = 0.997

19. E X : L ISTERIA MONOCYTOGENES pHT°C Aw = 0.95 pH T°C Aw = 0.93 µ = µ opt.  (T°, pH, aw)

20. G ROWTH B OUNDARY A UGUSTIN ET AL 2005 X : are environment factors (pH, T, aw…) C : are inhibitor factors concentration such as organic acids

21. L AG TIME MODELS Lag time for a microorganism depend on Environment parameter Physiological stage of the microorganism Relative lag time (RLT) RLT=lag time/generation time RLT=lag time x Ln2/Growth rate

22. P OPULATION VS C ELLULAR LAG TIME Growth rate of populations and single cells do not differ Lag time for populations (> 10-100 cfu/g) are shorter and less variable than for single cells Lag time of single cells (corresponding to contamination of some foods) can be predicted from population based data of similar physiological condition lag time x μmax = 3.9 ± 2.5 (single cells) Ln(Mean lagpopulation) = 0.907 x Ln(Mean lag single cell) – 0.311 Single cell lag is about two times the population lag

23. D ISTRIBUTION OF CELL LAG : OSMOTIC STRESS (N A C L 25% FO 24 H ) Cellular lag time variability

24. D ISTRIBUTION OF CELL LAG : HEAT STRESS (55°C FOR 4 MIN ) Cellular lag time variability

25. E VALUATION AND V ALIDATION Secondary models can be evaluated by comparing measured and predict values of kinetics parameters In real world the environment conditions vary during time, the experiment should be deigned to allow the combination of secondary and primary model Measurement of the organism concentrations And all the relevant physical and chemical parameters during time

26. S IMPLIFICATION ?  The model should take into the account for the food complexity!  Example Listeria monocytogenes in smoked fish Ross & Dalgaard 2004, Mejlholm & Dalgaard 2007

27. M ICROBIAL GROWTH MODELING Processing conditions Product characteristics Storage conditions ShelflifeCritical concentration of spoilage micro-organisms Safe shel-life Critiacal concentration of pathogenic micro-organisms Spoilage micro-organisms Pathogenic micro-organisms Storage time

28. 7 th Dubai International Food Safety Conference & IAFP’s 1 st Middle East Symposium on Food Safety MODELING BACTERIAL SURVIVAL OR INACTIVATION KINETICS

29. B ACTERIAL SURVIVAL OR INACTIVATION KINETICS “Survival curve” Same Micro-organism MediumTemperature Graph of the number of survivors according to time

30. S URVIVAL CURVE, INACTIVATION KINETICS t, time log 10 N 1 log 10 N 2 log 10 N D t1t1 t2t2

31. E QUATION if N 2 = N 1 /10, log 10 (N 1 /N 2 )= 1, t 1 – t 2 = time to divide the population by 10 = D slope = -1/D D =decimal reduction time D = decimal reduction time t, time log 10 N N N - 1 D

32. O THER WRITING log 10 (N) = log 10 (N 0 ) - t/D N = N 0 10 -t/D E = t/D = log 10 (N 0 /N) = « efficiency » = number of decimal reductions = number of log reductions = log kill

33. A N INTERESTING CONSEQUENCE N = N 0 10 -E Consider a lot of units of volume V, the expected number of survivors per unit is given by: N. V= N 0 V. 10 -E If N. V  1, then the unit is not sterile If N. V < 1, then the unit is sterile

34. S HOULDER timea log 10 N 0 log 10 N

35. S HOULDER Multi target theory e.g. clumps Multi hit theory Activation taking precedence over inactivation mechanism Cells loosing their resistance e.g. neutral spores in acid suspension medium e.g. inactivation of catalase

36. T AILING OFF timea log 10 N 0 log 10 N T AILING OFF

37. Mixed populations Clumping Activation of a secondary spore germination pathway Protective effect of the suspension medium e.g. acid spores in neutral medium

38. S- SHAPED CURVES Ababouch, L. et al., J. Appl. Bacteriol. 1987 62:503-11

39. G ENERALIZED EQUATION FOR EFFICIENCY timea a log 10 N 0 log 10 N

40. C ONCAVI TYUPWARD L'Haridon, R. & Cerf, O. Revue de l'Institut Pasteur de Lyon 1978 11: 445-456.

41. N ON LINEAR SURVIVAL CURVES 2/3 of experimental studies Many other equations can be used

42. I NFLUENCE OF TEMPERATURE B IGELOW T, temperature log 10 D n n - 1 z

43. temperature time Equal  ti T EMPERATURE CHANGES

44. M ODELING INDUSTRIAL TREATMENTS Each  t i achieves a number of decimal reductions E i = (  t i – a)/D Ti The total treatment achieves a total number of decimal reductions

45. M ODELING INDUSTRIAL TREATMENTS Pasteurizing value The F value for a process is the number of minutes required to kill a known population of microorganisms in a given food under specified condition Sterilizing value

46. E XAMPLE