integration of non-testing tools: a weight of evidence...
TRANSCRIPT
SETAC - October 24, 2008
Integration of non-testing tools:
a weight of evidence approach
Dinant Kroese
TNO Quality of Life
Zeist, The Netherlands
SETAC - October 24, 2008
Generation of information under REACH
Gathering all existing data
Data sharing
Animal studies only as a last resort
SETAC - October 24, 2008
Tonnage Human Health
1 – 10 tpa
Annex VII
• In vitro skin and eye irritation
• Skin sensitization
• In vitro mutagenicity
• Acute toxicity (one route)
10 – 100 tpa
Annex VIII
• In vivo skin and eye irritation
• Further in vitro mutagenicity
• Acute toxicity (2nd route)
• Sub acute toxicity (28d)
• Reproductive toxicity screen
100 – 1000 tpa
Annex IX
• Further mutagenicity tests
• Sub-chronic toxicity (90d)
• Reproductive toxicity tests
>1000 tpa
Annex X
• Further mutagenicity tests
• Further reproductive toxicity
tests
• Carcinogenicity may
• Chronic toxicity may
Required Information under REACHAnnexes VII - X
SETAC - October 24, 2008
REACH
Estimated number of substances
20,000substances
Annex VIII
≥ 10 tpa
Annex VII
≥ 1 tpa
Annex IX
≥ 100 tpa
Annex X
≥ 1000 tpa4,600
substances
2,900substances
2,600substances
SETAC - October 24, 2008
Scenario for animal use under REACH
2030
Anim
al use
Pre-registration
Dec 2008
Registration Annex X
+ CMR (cat 1+2) > 1 tonne/y
+ very toxic (R50/53) > 100 tonnes/y
Dec 2010
Registration Annex IX
June 2013
Registration Annex VII & VIII
June 2018
2008 2010 2020
SETAC - October 24, 2008
Different types of information
In vitro
(Q)SAR
Grouping & read across
Exposure
Human data Non-Testing information
SETAC - October 24, 2008
Combine information in Integrated Testing Strategies (ITS)
In vitro
Grouping & read across
(Q)SAR
Existing testing
information
Human data
Gathered information ..
via ITS
SETAC - October 24, 2008
Combine information in Integrated Testing Strategies (ITS)
Additional Testing Needed?
using WoE
cf Required information:Annexes VI-XI ?
In vitro
Grouping & read across
(Q)SAR
Existing testing
information
Human data
Perform the test!
SETAC - October 24, 2008
Combine information in Integrated Testing Strategies (ITS)
Additional Testing Needed?
using WoE
cf Required information:Annexes VI-XI ?
In vitro
Grouping & read across
(Q)SAR
Existing testing
information
Human data
Perform the test!
Waiving cf Annex XI:
Technical or Exposure arguments
?
SETAC - October 24, 2008
…taking into account:
Quality descriptors
Relevance of data
Acceptability in the field
Fit for purpose
Cost effectiveness
Pragmatism
Deadlines
Weight of Evidence
Making decisions on information in ITS ….
SETAC - October 24, 2008
REACH objectives
C&L, RC
When is information sufficient?
OK ???
Non-GLP / Non-OECD
Available information
+ =
+
+
SETAC - October 24, 2008
Evaluating information
• allows for transparency and objectivity
• assesses value of individual test & non-test information
• combines values of test & non-test information, expert judgement,
historical context
• reflects hierarchy of the testing strategy
• quantifies uncertainty
• resolves conflicting results….
a formal approach needed!
……that:
SETAC - October 24, 2008
Evaluating information
• allows for transparency and objectivity
• assesses value of individual test & non-test information
• combines values of test & non-test information, expert judgement,
historical context
• reflects hierarchy of the testing strategy
• quantifies uncertainty
• resolves conflicting results….
a formal approach needed!
……that:
SETAC - October 24, 2008
e.g. use of (Q)SARs
Human health effect
Human exposure to substance A
Animal model
SETAC - October 24, 2008
e.g. use of (Q)SARs
Mechanism A
Mechanism B
Mechanism C
Animal model
‘Same’ effect
SETAC - October 24, 2008
e.g. use of (Q)SARs
Mechanism A
Mechanism B
Mechanism C
Animal model
‘Same’ effect
(Q)SAR model, critical:
1. Understanding of toxicity mechanisms
biological domain
2. Specifying of applicability domains
chemical domain
Statistical model
SETAC - October 24, 2008
e.g. use of (Q)SARs
Animal model Statistical model
Uncertainty is defined by
variability sensitivityspecificity
SETAC - October 24, 2008
Evaluating information in ITS
ITS can be modeled as Decision Networks through
Influence Diagrams
Influence Diagrams are graphical networks with 3 types of
nodes, that model:
1. Chance events, e.g. probability of a chemical having a
certain property
2. Decisions, such as exposure-based waiving, decisions to
do more testing
3. Utilities: costs, animal use, value of information
using ‘Decision Analysis’
SETAC - October 24, 2008
Influence Diagram:
Does a compound have a certain toxicity?
Result Model A
Decide on running one or more non-testing models from toolbox
Costs, value, animal use
Result Model B Result Model C
Result Model I
Decide on running one or more testing models from toolbox
Result Model II
Mechanisms, Domain, Specificity, Sensitivity …
Result Model III
Sufficient certainty to make decision?
?
Dependence…
Predictivity…
SETAC - October 24, 2008
Landscaping document: conclusions
Webtool: position & functionalities
Route to webtool: concept deliverable
Takes time to achieve common understanding and agreement!
…
interaction between:toxicologist, chemists, statisticians, hygienists, toolbuilders…..
‘own’ language and concepts etc ….
SETAC - October 24, 2008EU Working Group on QSARs - OSIRIS
OSIRIS webtool features
Implements the ITS developed within RIP/OSIRIS
Compatibility of data formats from external sources
Is secure and keeps user privacy
ITS processes are persistent
ITS is easily changed
Is easy to use
SETAC - October 24, 2008
General concept of OSIRIS tool
WP 4.2 Tool
sorts to type of info
adds weight to type of info
or
identifies datagap
or
asks expert input
makes an assessment
compares with ‘golden standard’
information, and concludes on:
C & L
RA
concludes on additional
information needed
consults ITS and library of options
ITS library of
options
webtool
Input
Output
Predefined questionaire?
SETAC - October 24, 2008
General concept of OSIRIS tool
WP 4.2 Tool
sorts to type of info
adds weight to type of info
or
identifies datagap
or
asks expert input
makes an assessment
compares with ‘golden standard’
information, and concludes on:
C & L
RA
concludes on additional
information needed
consults ITS and library of options
ITS library of
options
webtool
Input Generate new test data
Predefined questionaire?
SETAC - October 24, 2008EU Working Group on QSARs - OSIRIS
OSIRIS webtool features – input & output
Input
Only public databases are used
Accepts testing data introduced by the end user
Testing data introduced by a specific user are not used in the
process of another user
Allows to include expert judgment in selected phases
Output
A report describing the whole process
It indicates what (testing) data is required to satisfy information
requirements
Can be tuned to minimize costs or animal use
SETAC - October 24, 2008
Concluding remarks
• WoE: yet no prototype……..
• Assesses and ‘adding’ weight to information is being worked on
• Positioning webtool in very dynamic field needs attention
• Very first webtool pilot on mutagenicity; only testing sofar
• A long way to go… there is an urgent need