The Spiced project: Seasonable doubt...

Saskia van Ruth (saskia.vanruth@wur.nl)

Partners Spiced WP4: BfR, BIOR, Fuchs, DLO-RIKILT

Are we dining on deception?

3

Souls for sale?

Food product adulteration and counterfeiting is a thriving multi-billion euro global industry. It is highly profitable and the risks of significant legal consequences are low.

Fraud reports in three databases

0 50 100 150 200

Herbs and spices

Olive oil

Fish and fish products

Milk and milk products

Meat and edible offal

Vegetable fats and oils…

Nuts, nut products and…

Honey and royal jelly

Other seafood…

Fraud cases 2008 - 2013

Food Fraud Food Shield RASFF

Food Fraud

Food Defence

Food Safety

Food Quality

Ideologically driven Motivation is ‘HARM’

Food borne illness

Economically driven

Motivation is ‘GAIN’ Intentional Adulteration

Unintentional / Accidental Contamination

Petra Wissenburg, Danone Corporate Quality & Food Safety

EU project SPICED fraud detection methods

The aim?

• To develop rapid and cost-efficient methodologies for detection of contamination (adulteration) of spices/herbs with chemical agents

• Aiming at early detection of anomalies in view of food integrity, i.e. signalling/flagging for unusual samples (authentic vs adulterated)

By which means?

• By exploring spectroscopic and spectrometric fingerprinting techniques

9

Development of broad anomaly fingerprint tests

0

20

40

60

80

100

120

140

160

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

0

20

40

60

80

100

120

140

160

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

10

Overview of activities

11

Method set-up

for spices

Data genuine samples

Data self-adul-

terated samples

Data

bases spices

Broad Anomaly models/ compa-

rison

Herbs

• Basil

• Thyme

• Oregano

Development of techniques for broad anomaly testing of spices

Spices

• Pepper

• Chili paprika

• Nutmeg

12

Oregano and product foreign material: leaves

Nutmeg and product own material/removal of valuable fractions

Paprika powder and unapproved ‘enhancements’: colourants

15

Selected analytical techniques

Spectroscopy

• FTIR

• NMR

Mass spectrometry

• PTR-MS

• DIMS

• (LC)-HRMS

• ICP-MS

16

Approach: one class classification model

Normal variation, Database samples (50)

Samples with deviations from what is considered normal

Building the best statistical model for discrimination

Future samples to be tested

17

Approach

50 Authentic baseline samples of each spice were analysed with each method

One class classification models were built based on those authentic samples

Inferior/adulterant material, and mixtures of authentic and inferior/adulterant material were analysed

The models were challenged with the authentic and modified sample material

Optimised classification models of the various methods were compared to select most promising approaches

18

Will the

real nutmeg please stand up?

Proton Transfer Reaction Mass Spectrometry

Nutmeg

1

10

100

1000

10000

100000

22

32

42

52

62

72

82

92

10

2

11

2

12

2

13

2

14

2

15

2

16

2

17

2

18

2

19

2

20

2

21

2

Inte

nsi

ty`[

pp

bv]

Mass [amu]

Nutmeg spectrum Terpenes: pinenes, carenes, limonene Myrcene, cymene, sabinene, camphene

1,2

-cin

eole

safr

ole

meth

yle

ugenol

asaro

ne

myristicin

acetic a

cid

1-b

uta

nol

One class classification result: Nutmeg

Baseline

sam

ple

s

Adultera

tions

Challenge

bla

nks

Cla

ss d

ista

nce

• SIMCA model, 7 factors • Normalised, 10log scaled

data

Classification results:

Prediction Authentic Adulterated % correct

Baseline samples 47 3 94% Challenge samples:

Authentic 5 0 100%

Adulterated 4 38 90%

Best models: discrimination 100% real nutmeg and own material mixed samples

Volatiles PTR-MS: • kNN model

• 100% real nutmeg and 100% correct prediction for own material mixed samples

• Estimated limit of detection 7%

Non-volatiles FI-ESI-MS: • PCA Q values

• Similar success rate predictions

• Limit of detection 30%

Successful classification PTR-MS - nutmeg

0

20

40

60

80

100

Authentic training Authenticchallenge

Adulterantchallenge

Mixed adulterantchallenge

[%]

24

Conclusions

Broad anomaly testing approaches have been developed for a variety of spices and herbs and various fraud issues in Spiced

For each spice/herb at least a few approaches were able to distinguish well between authentic and adulterated materials

Because of the one class classifier approach, the methods are expected to be able to detect a variety of other deviations from normal, authentic sample material

To increase robustness the training sets of authentic material could be expanded to cover more “special but natural” variation of the spice and herb materials

25

More presentations on the topic to follow

Bettina Horn (BfR): More fingerprinting methods, especially on paprika powders

Vadims Bartkevičs: The targeted approach

26

The partners

27

Thank you for your attention