DECISION FRAMEWORKS AND DATA INTEGRATION
Alternatives to Animals for Ecotox Testing Committee
Eco‐TTC Framework
Scott Belanger, PhD, P&G
Frameworks for intelligent non‐animal testing Framework development Alan Boobis, PhD, Imperial College Scott Belanger
HESI Annual Meeting
7 June 2016
•
Protection target: all possible species in all types of ecosystems
–
–
–
–
8.7 million described species
31,000 fish species alone
FW, SW, sediment, agricultural land, WWTPs
True goal: protection of ecosystem services
•
•
•
Energy flow
Nutrient flow
Biodiversity
•
Amount of available data
•
Often means product level exposure versus summed exposures from all
uses of a chemical, in other words total industry volumes are necessary
for assessments and understanding effluents which aggregate many
sources
•
Exceptional reliance on biostatistics and extrapolation
Tier I: a) phys-chem analyses, literature search,
read across
b) QSARs
REPLACEMENT
Tier II: non vertebrate or In vitro assays to
evaluate toxicity
Tier III: Refined In vivo tests to measure
toxicity
Tier IV: in vivo tests

Established/useful concept in human health
assessment
Noncancer
Endpoints
0.9 mg/kg/d
0.15 mg/kg/d
3 mg/kg/d
5th %ile
Screening-level hazard assessments
•
•
•
•
•
•
Maximizes resource use (animals, time, $$)
Potential for rapid-decision making
Fully utilizes existing knowledge
Allows evaluation of chemicals with little or no
toxicity data
Provides conservative estimates for low-production
chemicals
Supports read-across
Why now?
•
•
•
Toxicology data is abundant and high-quality
Increasing regulatory assessment needs
Improved ability to probe complex chemical
information / data







All existing (non-human) species
All environmental compartments
Huge diversity of data
Scattering of data sources
Changes in methods over time
Differences in regulatory schemes
AMOUNT OF DATA
WHAT
HOW
Large, diverse datasets
Ecotox Data Collection
Organization, data quality,
architecture
Data Harmonization / Curation
Assign to MOA, function, use,
etc.
Classify as acute / chronic
Data Characterization / Classification
‘SIFT’ Process
Access database development
MOA classification tools
Chemical classification tools
Acute / chronic logic tool
EcoTTC
DATABASE
Data analysis tools to derive
distributions
Distribution Metric
PNEC derivation tool
Data Analysis / ecoTTC
Derivation
R tool for probability
distributions
Threshold calculations
Expert analysis
Application directions
Database contents
USEPA
ECOTOX
68716 records
1864 substances
955 species
SSD db
29903 records
3447 substances
1557 species
Pharma
db
334 records
163 substances
3 taxa
ECHA
2398 records
215 substances
131 species
109,500 curated records
6200 unique CAS IDs
1900 species
12 original databases
Data Harmonization
/ curation
137,000
(> 1300 PT/Substances)
416
StepNotifiers
0: Purpose
Step 1: Relevance
Step 2: Validity
Step 3: Acceptance
200,000
Aquatic tox
3 trophic levels
158,800
Species, endpoint
132,000
CAS ID, chemical
concentration, qualifiers
120,300
Effect, duration
109,500
trust
transparent
Pool of data
meeting all criteria
ready for further curation
Adapted from Beasley et al., 2015. Environ Toxicol Chem 34,
1436-1442
US
Phys-chem/MOA
For each chemical
Not identified as a pro forma PNEC derivation method by any regulatory body
Europe
Page 1
Eco‐TTC flow: Identify most relevant PNECs for evaluation based on lowest trophic level available approach and geometric mean per trophic level approach
Page 1
ERA flow: Identify most relevant PNEC based on data quality and quantity
Japan
To Japan Flow Chart
Assignment of appropriate chemical category
Eco‐TTC not available; Use QSAR
Chronic data
Calculate geometric mean per species
Calculate geometric mean per species
Local QSAR available (e.g., class‐specific often based on homologue distribution)
Generic QSAR available
(e.g., ECOSAR, Danish QSAR, OECD Toolbox, etc.)
Empirical ecotoxicity data available (any species)
Acute data
PNEC1 based on 5%ile of distribution
Calculate geometric mean per trophic level
AF of 10,000 to
lowest trophic level result
AF of 10,000 to
lowest trophic level result
PNEC2
PNEC3
Enter eco‐TTC generation process
Identify lowest acute value per trophic level (lowest algal species, lowest invertebrate species, lowest fish species)
Enter eco‐TTC generation process
Calculate geometric mean per trophic level
Enter eco‐TTC generation process
Enter ERA PNEC determination process
Identify lowest chronic value per trophic level (lowest algal species, lowest invertebrate species, lowest fish species)
See next page, Eco‐TTC flow
Next page for Eco‐TTC flow
See next page, ERA flow
Next page for Eco‐TTC flow
Enter eco‐TTC generation process
Enter ERA PNEC determination process
AF of 1000
2 trophic levels present
AF of 1000 on most sensitive
AF of 100 on most sensitive
Acute and chronic toxicity
Is the taxon known to be slow growing?
Eurkaryotic, nonphotosynthetic stage microbes or strictly non‐
pohotosynthetic
multicellular organisms (e.g., Tetrahymena)
Do not consider further here; go to microinvertebrate
process
Prokaryotic nonphotosynthetic
microbes (e.g., Pseudomonas, Vibrio, Microtox)
Do not consider further (excluded)
Acute Toxicity:
NOECs for acute toxicity are not to be used
Chronic Toxicity:
Use lowest ECx
that is biologically meaningful (priority); use lowest NOEC if ECx
cannot be used
Acute Toxicity:
Endpoint expressed as duration and EC50 (boundary of EC30 to EC70)
Chronic Toxicity:
Endpoint expressed as (in order of most to least preferred) duration and EC10 (boundary of EC5 to EC20), NOEC, Chronic Value Do not consider further (exclude)
Terminal cell density or biomass sensu USEPA 850.4500
Acute test on missing trophic level is available
3 chronic trophic levels are available, including most sensitive acute taxon
AF of 5000 on most sensitive
PNEC9
AF of 1000 on most sensitive
10 or more species are available‐
perform SSD
If algae is missing taxon use ACR of 20
Apply ACR of 100 to amines
10 or more species are available‐
perform SSD
AF of 1 to 5
AF of 1 to 5
PNEC19
PNEC20
3 trophic levels present acutely, 1 chronic available (fish or Daphnia only)
Mesocosm or microcosm available
Apply AF of 5 to lowest available chronic data
Acute tests on both missing trophic levels are available
Apply ACR to additional acute data
If Daphnia is missing taxon use chemical specific ACR
If fish is missing data use ACR of 10
Apply ACR of 10 to non‐
amines
For algae use ACR of 20
Apply ACR of 100 to amines
<24 hrs
24‐96 hrs for most species; up to 10‐14 d, depending on culture, testing system
Do not consider further (exclude)
Include for further consideration
<24 hrs
Do not consider further (exclude)
24‐96 hrs for most species
Include for further consideration
Physiological or biochemical only What endpoint was measured?
Photosynthesis, biomass (pigments, cell density, biomass)
Photosynthesis rate at test end
Cell density and biomass surrogates) converted into growth rate sensu OECD 201 (Er, Eb)
What group is being assessed?
Apply ACR of 10 to non‐
amines
1 or 2 acute toxicity studies on trophic levels available
Apply ACR to acute data
Apply ACR to acute data
For Daphnia use chemical specific ACR
For fish use ACR of 10
For Daphnia use chemical specific ACR
For algae use ACR of 20
Apply ACR of 10 to non‐
amines
Apply ACR of 100 to amines
Apply AF of 10
AF of 10 to lowest value
For fish use ACR of 10
Apply ACR of 10 to non‐
amines
Apply AF of 10
Apply AF of 10
PNEC24
AF of 10 on lowest value
Identify lowest of the 3 trophic level values
AF of 10 on lowest value
AF of 100
AF of 10 to lowest value
Identify lowest of the 3 trophic level values
PNEC25
AF of 1 to 5
AF of 1 to 5
PNEC11
PNEC12
PNEC22
PNEC16
Larger invertebrates not covered in the flow diagram for microinvertwbrates
(including rotifers, Tetrahymena, Parameciums, and other ciliates)
3 acute toxicity studies on all trophic levels available
Apply ACR of 100 to amines
If fish is missing data use ACR of 10
If Daphnia is missing taxon use chemical specific ACR
If algae is missing taxon use ACR of 20
Mesocosm or microcosm available
PNEC15
PNEC10
Apply AF of 5 to lowest available chronic data
Apply ACR to additional acute data
PNEC18
AF of 10
3 trophic levels present acutely, 1 chronic available
1 chronic trophic level is available
PNEC17
2 trophic levels present
3 trophic levels present
3 chronic trophic levels are available, including most sensitive taxon
2 chronic trophic levels available
PNEC21
PNEC23
Macroinvertebrates
Yes
No
AF of 50
PNEC13
PNEC14
PNEC6
Microbes
Duration of the study
2 chronic trophic levels are available, including most sensitive taxon
AF of 10
3 trophic levels present
PNEC7
Prokaryotic photosynthetic organisms (Cyanobacteria, blue‐
green algae)?
PNEC8
AF of 10 on most sensitive
AF of 10,000
AF of 10
PNEC5
AF of 10
Eukaryotic photosynthetic organisms (algae)?
2 chronic trophic levels are available, including most sensitive acute taxon
1 trophic level present
1 chronic trophic level is available on less sensitive acute taxon
AF of 100
PNEC4
3 chronic toxicity trophic levels available
Europe
1 trophic level present
1 chronic trophic level is available on most sensitive acute taxon
What group is being assessed?
From geometric mean determinations (Page 2)
Japan
US
Not in EU TGD (2008), P&G guidance
Eco‐TTC available
ERA Processes
Page 3
Assess physical chemistry (minimum)
‐log Kow
‐solubility
‐pKa
‐MW (MWn, MWm)
‐distributions of homologous components
Do not consider further (exclude)
<24 hrs
Duration of the study
>24 to 7 days depending on the species, culture, and testing system
7 days and longer depending on the species, culture, and testing system
Do not consider further (exclude)
Acute Toxicity:
NOECs for acute toxicity are not to be used
Chronic Toxicity:
Use lowest ECx that is biologically meaningful (priority); use lowest NOEC if ECx cannot be used
Candidate for acute toxicity; common taxa will include Ceriodaphnia species, Daphnia species, Chironomus species, and various amphipods
Candidate for chronic toxicity; large array of species including those mentioned above
Physiological, biochemical or genomic only Acute Toxicity:
Endpoint expressed as duration and EC50 or LC50
(boundary of EC30 to EC70)
Survival (or immobility sensu
daphnid test guidelines)
Chronic Toxicity:
Endpoint expressed as (in order of most to least preferred) duration and EC10 (boundary of EC5 to EC20), NOEC, Chronic Value Survival, biomass, growth, organism density, reproduction endpoints, emergence (insects)
Chronic Toxicity:
Endpoints that are not assessed using a statistical model are not to be used
What endpoint(s) was measured?
Fish
Do not consider further (exclude)
<48 hrs
What group is being assessed?
Fish, broadly defined as both cartilaginous and bony fishes
Duration of the study
Freshwater petramyzontidae
(lampreys) are in scope
Sharks and rays (Chondricthyes) should be evaluated on a case‐by‐case basis
48 hrs to 7 days depending on the species, culture, and testing system
Candidate for chronic toxicity; large array of species
7 days and longer depending on the species, culture, and testing system
Do not consider further (exclude)
Candidate for acute toxicity; many commonly available test guidelines provide important context (OECD 203, USEPA 850 series, ASTM 1241)
Physiological, biochemical or genomic only Acute Toxicity:
NOECs for acute toxicity are not to be used
Acute Toxicity:
Endpoint expressed as duration and LC50
(boundary of LC30 to LC70)
Survival
Chronic Toxicity:
Use lowest ECx that is biologically meaningful (priority); use lowest NOEC if ECx cannot be used
Chronic Toxicity:
Endpoint expressed as (in order of most to least preferred) duration and EC10 (boundary of EC5 to EC20), NOEC, Chronic Value Survival, biomass, organism density, growth , reproduction, and deformity endpoints.
Chronic Toxicity:
Endpoints that are not assessed using a statistical model are not to be used
What endpoint(s) was measured?
Data Characterization: Mode of action
ECOSAR chemical classifications
from the evaluated dataset
(n=5636 unique CAS)
Verhaar MoA chemical classifications
(n=5636 unique CAS)
WHAT
HOW
Large, diverse datasets
Ecotox Data Collection
Organization, data quality,
architecture
Data Harmonization / Curation
Assign to MOA, function, use, etc.
Classify as acute / chronic
Data analysis tools to derive
distributions
‘SIFT’ Process
Access database development
MOA classification tools
Data Characterization / Classification
Chemical classification tools
EcoTTC
DATABASE
SQL database
Web-based user interface
Distribution Metric
PNEC derivation tool
Data Analysis / ecoTTC
Derivation
R tool for probability
distributions
Threshold calculations
Expert analysis
Application directions
Acute / chronic logic tool
Need for eco
TTC
Database
PNEC
derivation tool
Chemical
classification
•Description of database structure
•Sources of data
•Filtering process
•Analysis of database content
•Codify the options that can be used to evaluate ecotoxicity data using
ecoTTC database
•Communicate differences/similarities in regulatory outcomes
•Review of MOA assignment schemes
•Discussed tiered approach based on information
•Potentially describe what we used in ecoTTC databases (which are “most
appropriate” to use”)
A new approach….



eco TTC database and associated tools
publicly available
Workshop participants invited to develop case
studies to evaluate the approach, database,
and tools prior to the workshop
Results from various stakeholder exercises
will be presented and discussed