fighting unknown chemicals: analytical strategies for risk ... · fighting unknown chemicals:...

Fighting unknown chemicals: analytical strategies for risk prioritization

Eelco Nicolaas Pieke

Research Group for Analytical Food ChemistryNational Food Institute of Denmark

Berlin, Germany30 November 2017

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

In this talk:

• Introduction to IAS & NIAS: the chemical headache in foodo Knowledge gapso The analytical “Pillars of Knowledge”o Difference between IAS and NIASo Risk-assessing IAS and NIAS

• Proposed methods for closing the knowledge gap (not detailed)o Quantification (how much?)o Identification (what?)o Hazard character (how bad?)o Preliminary risk (priority?)

2

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Knowledge gaps in Risk Assessment

3

Intentionalor not?

Chemicalidentity

Knowledge on chemical intake

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Knowledge gaps in Risk Assessment

4

Intentionalor not?

Chemicalidentity

Knowledge on chemical intake

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Introducing the Pillars

5

[Identity & Potential] + [Quantity]

[Toxicity] + [Identity & Potential]

[Toxicity] + [Quantity]

= Exposure

= Hazard

= Effect level

Risk = Exposure + Hazard + Effect level

Identity &Potential

RISK

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

• Intentionally Added Substances (IAS):

– Limited number

– Pillars OK (well-defined data)

– Validated methods

– Substances are regulated

Existing tools & knowledge

Rarely problematic

A story of IAS and NIAS

6

?

IAS

Che

mic

als

in foo

d

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

• Non-Intentionally Added Substances (NIAS):

– Virtually unlimited number

– Pillars NOT OK (missing data)

– No formalized methodology

– No effective regulation

Lack of tools & knowledge

Unknown & undefined risk!

A story of IAS and NIAS

7

Unknown NIAS

Known NIAS + IAS

Che

mic

als

in foo

d

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Risk Assessment

8

Suspect chemical

Exposure assessment

Hazard assessment

Risk Assessment

Existing tox. data Animal testing

Assay testing

Blood/urine tests

Consumption Surveys

Sample analysis

Structure assessment

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Risk Assessment for unknown NIAS

9

Unknown NIAS

Exposure assessment

Hazard assessment

Risk assessment

Existing tox. data

Animal testing

Assay testing

Blood/urine tests

Consumption Surveys

Sample analysis

Structure assessment

No chemical structure

No availableexposure data

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Making informed decisions without tools?

10

Because of inadequate tools, there is no choice

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

What is needed?

11

• Methods to explore what is present & any effect on human health

• Decisions on unknowns should be information-based

• Data must be available fast & nonexhaustive:

» time is precious;

» information is expensive.

(standard matching, animal tests, identification studies, …)

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Chemicals in the tip of the iceberg

12

IASKnown NIAS

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Chemicals in the bottom of the iceberg

13

Lots of

Unknown NIAS

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

New tools to explore unknown chemicals

14

Quantitative data

LC-QTOF-MSLC Liquid Chromatography QTOF Quadrupole x Time of Flight MS Mass Spectrometry

TCMTotal Migratable Content: The chemical portion of a sample that has potential to migrate to food.

Quantity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Semi-quantitative tools

15

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8

[C] ± 3-fold error

Pieke et al., Anal Chim Acta. 2017 vol. 975, pp. 30-41.DOI: 10.1016/j.aca.2017.03.054.

Semi-quantify

Marker

Analyte Analyte

Quantity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Semi-quantitative tools

16

0

2

4

6

8

10

12

14

16

18

20

1 51 101 151 201 251 301 351 401 451 501 551

Concentration (µmol/L) -> 1 sample (recycled cardboard)

-> 1 analytical run

-> No identification needed

-> 550+ semi-quantified compounds0

5

10

15

20 Max. Conc.

Min. Conc.

Features

Pieke et al., Anal Chim Acta. 2017 vol. 975, pp. 30-41.DOI: 10.1016/j.aca.2017.03.054.

Quantity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Exploring the world of unknown chemicals

17

Semi-quantitative data

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8

Conc.

Quantity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Exploring the world of unknown chemicals

18

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8

Conc.

Chemical structure data Semi-quantitative data

Identity Quantity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Ambiguous structure assessment

19

Controlled FragmentationSimilarity Correlations

?First moleculein DB1

Second molecule in DB1

Third molecule in DB3

Database #1

First molecule in DB2

Second molecule in DB2

Database #2?

Pieke et al. (in press), J. Mass Spectrom.

Outputs

Identity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Ambiguous structure assessment

20

• Result: best-matching similar compound from available databases:

Identity

Pieke et al. (in press), J. Mass Spectrom.

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

The Three Pillars: incomplete

21

[Quantity] Semi-quantification

[Identity] Structure prediction

[Toxicity] ???

Identity &Potential

RISK

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Developmental Toxicity

The toxicity of tentative chemicals?

22

Carcinogenicity

Mutagenicity

Cramer Classification

???Q SAR–

Computer-based toxicity predictions

Model evaluation

Toxicity

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

The Three Pillars: nonexact data

23

Risk = Hazard * Exposure

Placeholders…

Identity &Potential

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

• We may not be able to calculate the true risk from approximated data, but …

• We can still get a nonconclusive perception of risk: preliminary risk assessment

Risk: starting from tentative data

24

Estimated Hazard Exposure

Risk Likeliness

Toxicity (QSAR) +Identity (Pred.) =

Quantity (Semi-Q) +Identity (Pred.) =

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Risk: a classification method

25

Rule-based decisions

Expertise decisions

Are exposure thresholds exceeded?”

Is the presence of this compound acceptable?

Classification

Identity &Potential

RISK

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Risk: the first things first

26

Identification

QSAR alerts

Exposure

FACTORS

Decision

A: Substances of direct concern

B: Substances of lesser concern

C: Insufficient or incomplete data

CLASSIFICATION

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

The purpose of tentative data

27

We needtentative data

… because the alternative isno data!

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

The purpose of tentative data

28

The choice to do nothing should be deliberate

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Thank you for your attention

29

Thanks to:

• Kit Granby• Jørn Smedsgaard

• Gilles Riviere• Bruno Teste

Eelco Nicolaas Pieke enpi(at)food.dtu.dk

• Elena Boriani

DTU Food, Technical University of Denmark Berlin, Germany 30 November 2017

Wish-list for NIAS screening

• Better identification methods and tools – more knowledge on HRMS needed

• Greatly improved databases for possible IAS & NIAS – ideally from the source

• More comprehensive tools for hazard – QSAR is not mature and subject to discussion

• More work on sampling & semi-quantification – reduce the uncertainty in the assessment

30