Predictive microbiology and risk assessment of non thermal technologies A. Martínez, F. Sampedro and D. Rodrigo Instituto de Agroquímica y tecnología de.

Predictive microbiology and risk assessment of non thermal technologies

A. Martínez, F. Sampedro and D. RodrigoInstituto de Agroquímica y tecnología de Alimentos (CSIC)

INNOVATIVE APPLICATION OF NON-THERMAL TECHNOLOGIES IN FOODS:

TECHNOLOGY, SAFETY AND CONSUMER ACCEPTABILITY

NOVEL PRESERVATION TECHNOLOGIES IN RELATION TO FOOD SAFETY

On account of a growing consumer demand towards foods that are safe, but retain the characteristics of fresh or freshly-prepared foods, mild preservation technologies are gaining more and more importance.

Examples include high-pressure processing, pulsed electric fields treatment, light technologies, cold plasma, and use of biopreservatives.


These mild preservation technologies enhance the shelf life of foods, are usually applied at room temperature and have a minor impact on the quality and fresh appearance of food products.

They are referred to as mild since they pose little stress on foods. This on the other hand increases the importance of food safety considerations.

Extended shelf life and a “fresh-like” product presentation emphasise the need to take full account of food safety risks, alongside possible health benefits to consumers.

Compared to a decade ago, research on mild preservation technologies has made a tremendous step forward. At this moment, novel technologies such as high pressure, pulsed electric fields and the use of biopreservatives are beyond the first development phase.

Equipment is available on different scales and several process conditions have been described.


The introduction of novel preservation methods stimulated research in microbiology, technology and food processing. The adoption of mild preservation technologies under European legislation is an ongoing process, as shown by the Novel Food Regulation. (Novel Food Regulation, Regulation No 258/97, 1997).



ModelingClassic log-linear survival model Modelling non log-linear survival curves Predictive modelling

FOCUS ON UNDERSTANDING MICROBIOLOGY

Behavior of microbial cells can be modeled in order to describe and/or predict microbial survival or growth.

Traditionally it was assumed that inactivation of microbial cells and spores exposed to heat or another stressing environment were governed by first-order reaction kinetics (in analogy with chemical reactions).

FOCUS ON UNDERSTANDING MICROBIOLOGY:MODELLING

The now classic log-linear method of D and z-values (Stumbo, 1973) is based on this theory and is widely accepted and practiced.

In this mechanistic model it is assumed that all cells have a similar resistance to the stressing factor and cell death occurs as a single critical event.


The D/z method has proven its value in food preservation for over 80 years, mainly in sterilization of canned products, but should we still use it?

Analysis of microbial inactivation resulting from these mild preservation methods revealed that many deviations from log-linearity occur.

These deviations can lead to over- or underprocessing when linearity is assumed and may result in a decrease in product quality or food spoilage respectively.


SigmoidalSigmoidalSC4SC4

TailTailSC3SC3

ShoulderShoulderSC2SC2

LinealLinealSC1SC1

Types of inactivation curvesTypes of inactivation curves


Simply assuming first-order kinetics of survival curves can therefore not be accepted anymore.

Non-lineal inactivation models:Non-lineal inactivation models:

))1(log(log))((

0

maxtBtk

BB eqqN

N

n

nnr

rtn

trtr

tr

n

rtB

1arctan

2arctanln

2

1)(

22

2

4 parameters

qB

kmax

r

n = 33 parameters

Baranyi model

)(

0

logMtBBM ee CeCe

N

N 3 parameters

C

B

M

Gompertz modified equation


Weibull frecuency distribution

b

0

taN

NLog

2 parameters

a

b



Some Examples

Inactivation of Enterobacter sakazakii by pulsed electric field in bufferedpeptone water and infant formula milk

M.C. Pina Pérez, D. Rodrigo Aliaga, C. Ferrer Bernat, M. Rodrigo Enguidanos, A. Martínez López (International Dairy Journal 17 (2007) 1441–1449)





Comparisson of fitting experimental data by Weibull distribution function (······), Gompertz modified equation () and Baranyi model ( ) for inactivation of E. coli cells at 150 (), 175 (), 200 () y 225 MPa ()

Inactivation of E. Coli and L. inocua by Hight Hydrostatic Pression in a vegetable beverage

Daniela Saucedo-Reyes, Dolores Rodrigo-Aliaga, Aurora Marco-Celdrán and Antonio Martínez-López (II Congreso Iberoamericano sobre seguridad alimentaria CIBSA 2008)

Comparisson of fitting experimental data by Weibull distribution function (······), Gompertz modified equation () and Baranyi model ( ) for inactivation of L. innocua at 200 (◊), 225 (□), 250 () y 275 MPa (○) MPa.



Goodness of fitting for the different models


Validation of models used in the study

Predictive modeling is regarded as the solution for microbial safety. These models can provide useful information on microbial behavior without experimental effort. Although they can be used within the limits of the model, extrapolations or true predictions are not always allowed or correct.

FOCUS ON UNDERSTANDING MICROBIOLOGY:PERDICTIVE MODELLING

In the future new tools should be developed to describe the potential hazards with a huge variety of products and formulations and suitable for expossure assessment studies. In these models a precise description of the behaviour of the mircoorganisms should be accounted for. However, the development of a precise or complete mechanistic model requires more information on the physiology of cell populations as well as single cell behavior.

FOCUS ON UNDERSTANDING MICROBIOLOGY:PERDICTIVE MODELLING


Novel food technologies such as high pressure and pulsed electric field treatments are used or will be used in the near feature for commercial applications.

Research is needed to evaluate the effects on food quality.

Issues such as safety and legislation have to be considered.


The Risk Analysis process

FOCUS ON IMPLEMENTATION:RISK ANALYSIS

ALOP

Risk Analysis

HACCP

GHP’s/ GMP’s/ GAP’s

FSO Food Safety Objective

Goverment Level

Industrial Level

FOCUS ON IMPLEMENTATION:RISK ANALYSIS

Risk assessment and management procedure, which follows the requirements of the Codex Alimentarius Commission, results in the final food safety objectives

Hazard identification

Hazard characterization

Exposure assessment

Risk characterization

FOCUS ON IMPLEMENTATION:RISK ASSESSMENT

FOCUS ON IMPLEMENTATION:EXPOSURE ASSESSMENT

An Example

The Monte Carlo simulation is used to establish the most influential parameters on the final load of pulsed electric fields E. coli cells

C. Ferrer, D. Rodrigo, M.C. Pina, G. Klein, M. Rodrigo, A. MartÍnez (Food Control 18 (2007) 934–938)

PEF TREATMENT OF A BLENDED ORANGE-CARROT JUICE INOCULATED WITH E. coli

Nomenclature: a scale parameter of the Weibull distribution function aE scale factor of the Weibull distribution at an electric field

strength (E) aR scale factor of the Weibull distribution at the referent

electric field strength (ER) N shape parameter of the Weibull distribution function E electric field strength (kV/cm) ER referent electric field strength (kV/cm) % percentage of carrot juice %R referent percentage of carrot juice z% variation on % of carrot producing a variation of ten fold

on scale parameter (a) zE variation on electric intensity strength producing a

variation of ten fold on scale parameter (a)


An example

N0: Initial microbial load (CFU/ml)

ZE: Electric field strength (kV/cm)

Z%: Juice percentage (%)

a: Kinetic parameter (s)

DEPENDENT VARIABLE (OUTPUTS)

INDEPENDENT VARIABLES (INPUTS):

N: Number of microorganisms after the process (CFU/ml)

FOCUS ON IMPLEMENTATION: EXPOSURE ASSESSMENT

An example


An example

Predicted aE values, as a function of electric field strength and carrot juice percentage, were obtained using the following predictive model:


An example

Ln(aE)=Ln(aR)-(B-C)/D-(E-ER)/F

The final number of microorganisms was predicted by using the following equation:


An example

Ln(S)=-(t/aE)n

1) Inputs were defined by distribution functions:

BestFit @Risk4.5 (Palisade Corporation)

a Beta General (0.38372; 0.32221; 7.9828;

16.1975)

NoNormal (12535984; 2488898)

ZEExtreme value (8.5689; 2.6152)

Z%Normal (27.4814; 4.5023)


An example

Inputs predicted values (a, N0 zE y z%)

N0 a zE z%

Experimental values

12535984 ±

2488897

12.7597 ±

2.9001

13.3467 ±

6.5801

23.4573 ±

1.6961

Monte Carlo Simulation

12535931 ±

2488749

12.4230 ±

3.1538

10.0784 ±

3.3530

27.4814 ±

4.5033

2) Monte Carlo Simulation:


An example


An example

No

%

E

-1 -0,5 0 0,5 1

1

No

E

%

-0,5 0 0,5 1

1

0% carrot juice

60 % carrot juice

40 µs, 25 KV/cm

At low intensities N0 is the most influent factor

3) Sensivity analysis


An example

E

No

%

-1 -0,5 0 0,5 1

1

%

E

No

-1 -0,5 0 0,5 1

1

0% carrot juice 60 % carrot juice

70 µs, 35 KV/cm


An example

At higher intensities E is the most influent factor

E

No

%

-1 -0,5 0 0,5 1

1

E

%

No

-1 -0,5 0 0,5 1

1

100 µs, 40 KV/cm


An example


Predictive microbiology and risk assessment of non thermal technologies A. Martínez, F. Sampedro and D. Rodrigo Instituto de Agroquímica y tecnología de.

Documents

Predictive microbiology and risk assessment of non thermal technologies A. Martínez, F. Sampedro and D. Rodrigo Instituto de Agroquímica y tecnología de.