Numerical Benchmarks for protecting biota against radiation in the environment Methodology to derive benchmarks, selected methods used in PROTECT and Results PROTECT FP6-036425

Download Report

Transcript Numerical Benchmarks for protecting biota against radiation in the environment Methodology to derive benchmarks, selected methods used in PROTECT and Results PROTECT FP6-036425

Slide 1

Numerical Benchmarks for
protecting biota against
radiation in the environment
Methodology to derive benchmarks, selected
methods used in PROTECT and Results

PROTECT

FP6-036425


Slide 2

Methods to derive benchmarks (reminder)

3 main methodologies :
(1) Deterministic : based on application of Assessment
Factor to critical ecotoxicity data, or
(2) Probabilistic : based on Species Sensitivity
Distribution (SSD) approach associated with an
arbitrary cut-off value, usually set at a protection level of
95% of the species, and the use of an additional AF,
(3) a weight of evidence approach using effect data from
field exposures
(1) & (2) based on critical ecotoxicity values from tests in lab
e.g., EC10 (preferred to NOEC) : stressor level in a given medium giving 10% effect in the
exposed group in comparison to the control group for chronic exposure.

(1) to (3) usually applied for chemical substances to derive PNEC (e.g., TGD, 2003)
Concerning RAS:

(1) and (2) applied and compared within ERICA to derive PNEDR (Garnier-laplace et al., 2006)
(1) used by Environment Canada (2003) for radionuclides
(3) used by Thompson et al. (2005) for multipollution context (radionuclides + metals)
PROTECT

FP6-036425


Slide 3

Methods used within PROTECT
• Following PROTECT Consultation and
availability of various EDR10 values (from ERICA and
follow up), SSD method adopted when data sets
were adequate (quality and quantity)
(1) allow to make the best use of the available most relevant knowledge in a
transparent way,
(2) allow to quantify the associated uncertainty,
(3) easily allow to revise the outcoming values when new knowledge
becomes available

• Need to incorporate documented expert
judgement at various stages
(1) when selecting ecotoxicity data (EDR10 in our case),
(2) when selecting a cut-off value to estimate the Hazardous Dose Rate of
interest (HDR5),
(3) when applying an AF to obtain the final benchmark (PNEDR).
PROTECT

FP6-036425


Slide 4

The 3-step method used
STEP 1 – Compiling quality assessed exposure-effect data
FREDERICA
Radiation Effect Database

Effect (%)

Data sorted per ecosystem, per exposure condition, per
bibliographic reference and per test . Quality of data describing each
test was assessed against 3 criteria (dosimetry, exp. Design, stats).
Multi-rules applied to accept or reject each data set

STEP 2 – Building dose rate-effect relationships to
estimate critical ecotoxicity values EDR10

100 %
50 %

Dose-effect relationship was built for each accepted test.
The quality of the fitted model was judged before accepted
EDR10. Two models tested (logistic/hormetic)

10 %
EDR50
EDR10

Dose Rate (µGy/h)

PAF (% of Affected Species)
100
80
60
40
Dose Rate (µGy/h)
20
0
1
10
100 1000 10000
PROTECT

STEP 3 – Deriving benchmarcks
Application of SSD on the estimated toxicity values accepted
after step 2.
Application or not of the taxonomic weighted options
Application of an AF to the HDR5 to obtain the final benchmark
FP6-036425


Slide 5

EDR10 from logistic models
Data set for one test
(a test is defined as a consistent group of (dose or dose rate, effect) couples from a given species and a given
effect, examined under defined exposure conditions (duration, irradiation pathway)

Logistic model

The data set is made of:
 at least 3 different couples (dose or dose rate, effect) including one for the control group (no dose(rate))
 at least two different couples if the effect is analysed relatively to the control.
YES

Effect

Control
(response at dose rate 0)

NO

data set is rejected

 The variation of effect with dose (or dose rate) is monotonous.
 The pattern is consistent with the state-of-the-art on the tested effect
YES

NO

data set is rejected

 The maximum effect value in case it was not reached during the test can be fixed theoretically if knowledge on such
effect is sufficient. The difference between the maximum effect value observed y(MaxObs) during the test and the
theoretical one y(MaxTheo) are used to calculate the extrapolation percentage needed to model dose-effect relationship
as follows : %Extrapol = 100 *(y(MaxObs) – y(MaxTheo))/(y(ControlObs) – y(Maxtheo)) where y(ControlObs) is the
effect value observed for the control group.
EDR10

Dose rate (µGy/h)
EDR10

YES

NO

data set is rejected

 At least one couple is located within the 10 to 90% of the variation of effect observed. This latter is defined as
(y(ControlObs)-y(MaxObs))
YES
NO
data set is rejected

 The Estimated ED50 or EDR10 are between two experimental couples.
YES

NO

data set is rejected

The Estimated ED50 or EDR10 can be used within a SSD analysis.

PROTECT

FP6-036425


Slide 6

EDR10 from hormetic models
Dose-response exhibiting an initial response stimulation or an effect minimisation

Curve shape

« Hormetic » model

NOEC definition

Inverted U shaped
curve

Selection criteria

Control
(response at dose rate 0)

NOEC definition

U shaped curve

EDR10

Exclusion criteria

Selection criteria

the highest dose with a response ≥ 90% of the control
- at least 5 dose-response data points (the minimal number to fit a
hormesis model with 4 parameters, requires fixing the lower limit to 0)
- 1 control data point
- at least 2 doses ≤ NOEC with a response numerically higher than the
control
- 1 NOEC
- at least 1 dose > NOEC with a response ≤ 90% of control
the highest dose with a response ≤ 110% of the control
- at least 6 dose-response data points (the minimal number to fit a
hormesis model with 5 parameters; lower and upper limit are different to
0)
- 1 control point
- at least 2 doses ≤ NOEC with a response numerically lower than the
control
- 1 NOEC
- at least 1 dose > NOEC with a response ≥ 110 % of control
(1) the absence of a relevant control;
(2) the incapacity to achieve responses greater than (or less than,
depending on end point) the control response (e.g., studies where the end
point was survival and the control response was 100% or where the end
point was tumour incidence and the control response was zero);
(3) at least two doses below the NOEC;
(4) at least one dose showing a priori criteria-based inhibition.

Dose rate (µGy/h)

PROTECT

FP6-036425


Slide 7

STEP 1 – EDR10 selection (1/4)
•ERICA : 82 EDR10 obtained for chronic g external exposure
meeting the previous criteria, covering 18 species (5 plants, 9
invertebrates, 10 vertebrates) and various effect endpoints
grouped as one geometric mean per effect category
(mortality, morbidity, reproduction) and per species (ie. for a
given species, a single value per category of effect (the GM)-> this allows to ignore
intra-species variation for the same effect category)

SSD on 24 GM -> HDR5=82 µGy/h [24-336]

•PROTECT : 104 EDR10 obtained for chronic g external
exposure meeting the previous criteria, covering 23 species (6
plants, 8 invertebrates, 9 vertebrates) and various effect
endpoints
PROTECT

FP6-036425


Slide 8

STEP 1 – EDR10 selection (2/4)
•Among those 104 EDR10,
•One single endpoint per species (the lowest value)
•Relevance for population sustainability but expert
discussion on ecological relevancy of some of them

•To illustrate the sensitivity of the SSD method and the derived
screening level dose rate, two sets of EDR10 selected:
« repro » list: most sensitive reproduction endpoint per
species (at the population level -i.e. include juvenile survival)
« ecol » list: most relevant with regard to « direct » ecological
interpretation

PROTECT

FP6-036425


Slide 9

STEP 1 – EDR10 selection (3/4)
« repro » list : 19 EDR10
ID

WildlifeGroup

Taxo

µGy/h

SpeciesL

EffectDescription

EDR10

Daphnia magna

Larval survival to starvation during 5 days, brood 1 (% survival when food lacks) - stress on stress
test of indirect effect on energy allocation to juveniles production

18537

Gilbin

Crustaceans

Invertebrates

1065

Crustaceans

Invertebrates

Daphnia pulex

Population birth rate (per day) Data read from graph

277633

247

Crustaceans

Invertebrates

Porcellio scaber

Mean number of offspring per tank per dose rate group

1030

Hertel-Aas

Annelid

Invertebrates

Eisenia foetida

Hatchlings per adult during the whole 13 weeks reproduction exposure period (F1/ Adult F0)

3369

296

Molluscs

Invertebrates

Mercenaria mercenaria

Survival of juvenile clams (%) on day 426. Dose = max. cumulative dose

119117

361

Annelid

Invertebrates

Ophryotrocha diadema

The percentage of worms in generation 3 surviving to day 62.

2359

326

Molluscs

Invertebrates

Physa heterostropha

Percentage of eggs that hatched

61229

523
841

Plants
Plants

Plants
Plants

Abies balsamea
Fagopyrum esculentum

Summary of mean fir characteristics for seven dose-rate catergories, Number of buds (1975),
Productivity in M3 generation (1979), Yield of seeds (g/sq m)

2945
X
40151

416
998

Plants
Plants

Plants
Plants

Pinus rigida
Triticum sinskjae

Effect of long term irradiation on seed development. Dose rate provided as average per day
Productive bush amount, % of the control value. N 13, K-48993 T. sinskjae var. Sinskajae

710
6434
X

998

Plants

Plants

Triticum monococcum

Productive bush amount, % of the control value. N 3, K-29602 T. monococcum var. Atriaristatum

8573
X

616
Egami
207

Mammals
Fish
Fish

Vertebrates
Vertebrates
Vertebrates

Mus musculus
Oryzias latipes
Pleuronectes platessa

Nº of litters per fertile female during 245 days (mean; SE).
Male mean gonad weight (mg)
Mean proportion of plaice testes occupied by different cell types irradiated for 197 days - sperm

26
19730
X
53
X

74
593
621
629

Fish
Mammals
Mammals
Mammals

Vertebrates
Vertebrates
Vertebrates
Vertebrates

Poecilia reticulata
Rattus norvegicus
Spanish goat
Sus crofa

Mean life time fecundity
A1 Spermatogonia ( % of control)
Total sperm production (% of control): 730 days irradiation
Testis weight (g) at 70 days of age (+- SE)

516
24
X
12
X
6.7
X

PROTECT

FP6-036425


Slide 10

STEP 1 – EDR10 selection (4/4)

« ecol » list : 14 EDR10
Omission of endpoints with difficult ecological interpretation (no direct
link with reproductive success) e.g., bud production, gonad weight,
µGy/h
sperm cell or spermagonia count
ID

WildlifeGroup

Taxo

SpeciesL

EffectDescription

EDR10

Gilbin

Crustaceans

Invertebrates

Daphnia magna

Larval survival to starvation during 5 days, brood 1 (% survival when food lacks) - stress on stress
test of indirect effect on energy allocation to juveniles production

18537

1065

Crustaceans

Invertebrates

Daphnia pulex

Population birth rate (per day) Data read from graph

277633

247

Crustaceans

Invertebrates

Porcellio scaber

Mean number of offspring per tank per dose rate group

1030

Hertel-Aas

Annelid

Invertebrates

Eisenia foetida

Hatchlings per adult during the whole 13 weeks reproduction exposure period (F1/ Adult F0)

3369

296

Molluscs

Invertebrates

Mercenaria mercenaria

Survival of juvenile clams (%) on day 426. Dose = max. cumulative dose

119117

361

Annelid

Invertebrates

Ophryotrocha diadema

The percentage of worms in generation 3 surviving to day 62.

2359

326
841

Molluscs
Plants

Invertebrates
Plants

Physa heterostropha
Fagopyrum esculentum

Percentage of eggs that hatched
Productivity in M3 generation (1979), Yield of seeds (g/sq m)

61229
40151

416
998

Plants
Plants

Plants
Plants

Pinus rigida
Triticum monococcum

Effect of long term irradiation on seed development. Dose rate provided as average per day
Fertility, % of the control value. N 15, K-23653 T. monococcum var. Hornemanni

710
9819

616

Mammals

Vertebrates

Mus musculus

Nº of litters per fertile female during 245 days (mean; SE).

26

74
622
204

Fish
Mammals
Fish

Vertebrates
Vertebrates
Vertebrates

Poecilia reticulata
Spanish goat
Oryzias latipes

Mean life time fecundity
Number born per female (in 5 consecutive gestation: see comment)
Number of surviving fish

516
1968
90861

PROTECT

FP6-036425


Slide 11

STEP 2 - SSD-fit to the selected list (1/3)
• Repro-list
SSWD - Log Normal

R² = 0.9583
KSpvalue = 0.500

Sp = unw eighted; TW: none

w m.lg = 3.24
w sd.lg = 1.41

100%

HDR5 = 8.4 µGy/h
CI 95% = [1.4;99]

Cumulative weighted probability

90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0,1

1

10

100

1000

10000

100000

1000000 10000000

Concentration
Dose
rate (µGy/h)

PROTECT

Best-Estimate

Centile 5%

Plants

Invertebrates

Centile 95%

Vertebrates

FP6-036425


Slide 12

STEP 2 - SSD-fit to the selected list (2/3)
• Ecol-list
SSWD - Log Normal

R² = 0.9625
KSpvalue = 0.500

w m.lg = 3.81
w sd.lg = 1.12

Sp = unw eighted; TW: none

HDR5 = 91 µGy/h
CI 95% = [11;710]

100%
Cumulative weighted probability

90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1

10

100

1000

10000

100000

1000000

10000000

Concentration
Dose
rate (µGy/h)

PROTECT

Best-Estimate

Centile 5%

Plants

Invertebrates

Centile 95%

Vertebrates

FP6-036425


Slide 13

STEP 2 - SSD-fit to the selected list (3/3)

Taxonomic weight options to fit SSD
Influence of the proportion of tested species from different taxonomic groups
(trophic levels) on the HDR5 while giving the same weight to each
trophic/taxonomic group
Repro
Ecol

Number of EDR10 Number of species Taxonomic Weight
19
19
No (5 P; 7 I; 7 V)
Yes (1 P; 1 I; 1 V)
14
14
No (3 P; 7 I; 4 V)
Yes (1 P; 1 I; 1 V)

HDR5
8.4
11
91
72

min
1.4
1.8
11
8.0

max
99
159
710
566

For the repro-list, applying the same weight for each TG means to decrease the
proportion of the most radiosensitive group from 37% (7/19) to 33% (equiprobability) ->
HDR5 increases slightly
Ecol-case : V from 28% to 33% -> HDR5 decreases
PROTECT

FP6-036425


Slide 14

STEP 3 - Screening generic dose rates
(in µGy/h)
Rounded to the nearest 10
Weight option

HDR5

min

max

AF

PNEDR

Generic Screening Value

Repro

No

8.4

1.4

99

1

8.4

10

Ecol

Yes
No

11
91

1.8
11

159
710

1
5

11
18

10
20

Yes

72

8.0

566

5

14

10

•AF= from 1 to 5 according to the TGD
- taxonomic coverage
- coverage of the inter-species sensitivity
- ecological relevancy of endpoints
- line of evidence from field
- final value compared to background
e.g., 0.1 to 6 µGy/h for marine organisms (Brown et al., 2004)
0.4 to 4 ________freshwater______________________
0.07 to 0.6______terrestrial________(Beresford et al. in press)

PROTECT

FP6-036425


Slide 15

STEP 3 - Screening generic dose rates
(in µGy/h)
Rounded to the nearest 10
Weight option

HDR5

min

max

AF

PNEDR

Generic Screening Value

Repro

No

8.4

1.4

99

1

8.4

10

Ecol

Yes
No

11
91

1.8
11

159
710

1
5

11
18

10
20

Yes

72

8.0

566

5

14

10

•For illustration purpose: AF=1 for the repro-list and 5 for the ecol-list

generic screening level recommended by PROTECT : 10 µGy/h

PROTECT

FP6-036425


Slide 16

STEP 3’ – Taxonomic screening dose rates
(in µGy/h)
•Data were insufficient to create an SSD for the 3 basic taxonomic groups (plants(5),
invertebrates(7), vertebrates(7))

•A limited sensitivity analyses of the overall SSD suggested that the only statistically
justifiable separation was to remove vertebrates leaving an SSD for plants and
invertebrates (with a total of 12 data points)
Weight option
Non-Vertebrates No (5 P;2 I-M;2 I-A; 3 I-C)
Yes (same weight per group)

HDR5

min

max

AF

PNEDR

Taxonomic Screening Value

449

166

1791

1

449

450

503

167

1858

1

503

500

•We suggest that a taxonomic screening level of 450 µGy/h be used for plants and
invertebrates in refined assessments if the results of the screening tier assessment
suggest this is required. For vertebrates, a taxonomic screening level of 10 µGy/h
(i.e. the generic screening level) is recommended.

PROTECT

FP6-036425


Slide 17

Regulatory action level
• The statistical extrapolation approach could be used to help in the
management decision making process by generating different levels
of potential impact (e.g. by taking the 50th percentile of the EDR10
distribution or 20th percentile of the EDR50 distribution etc).
• PROTECT will thus provide a range of tabulated effect levels and
percentiles which may prove to be useful to decision-makers
•Different percentiles of an unweighted SSD derived using EDRx values from the
repro-list as input and following the same derivation methodology as used for the
HDR5 values
Input data for SSD fit
Number of data
r2

HDR5
HDR10
HDR20
HDR50
PROTECT

EDR10

EDR25

EDR50

19

17

14

0.958

0.965

0.953

8
27
114
1741

30
85
302
3385

86
217
665
5641
FP6-036425


Slide 18

Summary of the proposed approach
and associated benchmarks (1/2)
•First tier of an ERA-type approach (simple – conservative)
The use of the proposed generic screening level should enable assessed sites for
which incremental dose rates are estimated as being below 10 µGy/h to be confidently
excluded from further assessment.
•Second tier of an ERA-type approach (less conservative)
The application of the 10 µGy/h screening dose rate for vertebrates
and 450 µGy/h for plants and invertebrates is likely to identify those sites which
pose an insignificant risk when used in such assessments.
•The generic and taxonomic screening values we have derived are within the range of
values suggested as being appropriate for population level protection by other
organisations using expert judgement.
•Suitable data we have identified are currently insufficient to enable the derivation of
more refined levels of taxonomic screening values (if these are required) using the SSD
approach.
PROTECT

FP6-036425


Slide 19

Summary of the proposed approach
and associated benchmarks (2/2)

•Regulatory action level
We have presented a SSD-based scientific analysis of the available data which
may help decision makers select a value appropriate to either planned or
existing situations in consultation with appropriate stakeholders.

PROTECT

FP6-036425