Chemical risk assessment: Chemical risk assessment: Historical perspectives and current trends Jean Jean Lou Lou Dorne, Dorne, European European Food Food Safety Safety Authority Authority, , Unit on contaminants in the Unit on contaminants in the food food chain chain
49
Embed
Chemical risk assessment:Chemical risk assessment ... · Chemical risk assessment:Chemical risk assessment: Historical ppperspectives and current trends ... 2005 Toxicol Sci 86, 20-26.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Chemical risk assessment:Chemical risk assessment: Historical perspectives and p p
current trends
Jean Jean Lou Lou Dorne, Dorne, ,,EuropeanEuropean Food Food SafetySafety AuthorityAuthority, ,
Unit on contaminants in the Unit on contaminants in the foodfood chainchain
Acknowledgments Acknowledgments
Authors
EFSA, Parma, ItalyDjien Liem , Bernard Bottex, Claudia Heppner
CCD, Center for statistics on AIDS Chiang Mai, ThailandBilly Amzal
Center for Ecology and Hydrology , UKDavid Spurgeon and Claus Svendsen
University of Southampton, Professor Andrew Renwick OBEProfessor Andrew Renwick OBE
Scientific panel on contaminants in the food chain of EFSA
OutlineOutline
• Historical perspectives and principles
• Population variability and risk assessment
• Systematic review, meta-analysis and chemical y yrisk assessment
• Conclusions
3
Historical perspectives and principlesHistorical perspectives and principles
Hazard identification and Exposure assessmentcharacterisation
Levels in food, feed, water, environmental media, dietaryexposure, food consumption,
ADME, acute/sub‐chronic/chronic toxicity, human data, genotox, reprotox, mode of action,p , p ,
relevant food groups, time trends
Deterministic vs probabilistic
human data, genotox, reprotox, mode of action, NOECs (ecological)NOELs, LOAELs (animal, human)mathematical modelling (BMD), Health based guidance (TDI…)
X vsg ( )
Risk characterisationRelate exposure to environmental standards (ecological),
margin of safety (animal)margin of safety (animal)TDI, derive Margin of Exposure (human)
GenotoxicGenotoxic carcinogenscarcinogens
Margin of Margin of exposureexposureMargin Of exposure (MOE) developed, by the JECFA and EFSA (2005) Point of reference on the dose-response curve* (based on animal and human data) divided by the estimated human intakes. MOE (animal data) >10,000 as of low concern for public health.
BMD/BMDL: Benchmark dose/ limit(*NOAEL: No observed-Adverse-Effect-Level)
Missing data, uncertainty in the database, subgroups, mixtures
“All things are toxic and there is nothing without poisonous qualities: it is only thewithout poisonous qualities: it is only the dose which makes something a poison”
PARACELSUS (1493-1541)
Pharmaco/Toxicokinetics Pharmaco/Toxicodynamics
What the body does to thechemicalHow the chemical is eliminated
What the chemical does to thebodyHow the chemical exerts itsfrom the body or activated into
a toxic species (ADME )
How the chemical exerts itspharmacological effect/ toxicityTarget receptor/cell/organ
10
EFSA‘s Risk assessment ofcoccidiostats: Cross-contaminationcoccidiostats: Cross contaminationof non-target feedingstuffs: Animalhealth and human health
Feed safety conference 6-7/10/2009 Wageningen
11
Coccidiostats in animal feeds
Ionophoric polyethers Non-ionophoric
O
CH3
HO
O
H3C
O
H3C H3CH3C
CH3
H
Cl
CNCl
Cl
N
N
N
O
O
OO
O
CH3
O
H3C
H3CO
OH
O
O
OH
OHH
H H H
Monensin ALasalocid
S li iSalinomycinNarasin
MaduramycinSemduramycin
DecoquinateDiclazuril
HalofuginoneNicarbazin (DNC/HDP)
R b idi
12
Robenidine
C idi t tCoccidiostatsHazard identification and characterisation
:Toxicological effectsExposure assessment
Occurrence in feed(Cross-contamination of
2% 5% and 10%)
:Toxicological effects
Toxicity in non-target animal
Occurrence/ residues in animal tissues/
Consumption in humansToxicity in laboratory
animalsToxicokinetics
/residues in 2%, 5% and 10%)
-Feed consumption in
non-target animal species
speciesLOAEL/NOAEL
p
LOAEL/NOAELanimal tissues
Uncertaintyfactor 100
Exposure in non-target animal species
TDI
Human exposurep
13
Risk characterisation in animals Risk characterisation in humans
Population variability and risk assessment
14
Use of uncertainty factors (UFs)Use of uncertainty factors (UFs)
SPECIESDIFFERENCES
HUMANVARIABILITY
10 10
KINETICS DYNAMICSKINETICS DYNAMICS
E t l ti f f t t i l tExtrapolation from group of test animals to average human and from average humans to potentially sensitive sub-populationssensitive sub-populations
Uncertainty factorsUncertainty factors
100 - FOLD UNCERTAINTY FACTOR
INTER-SPECIESDIFFERENCES
INTER-INDIVIDUALDIFFERENCES
Chemical specific adjustment factors can replace the
DIFFERENCES
10 - FOLDDIFFERENCES10 - FOLD
pdefault uncertainty factors (IPCS, 2006)Use of PB-TK, PB-
TOXICO-DYNAMIC
TOXICO-KINETIC
TOXICO-DYNAMIC
TOXICO-KINETIC
TK -TD models when data available for chemical of interest
10 0.4
2.510 0.6
4.010 0.5
3 210 0.5
3 24.0 3.2 3.2
Uncertainty Factors:Uncertainty Factors:TTowards owards a more flexible frameworka more flexible framework
Data-derived Data-derived
Toxicokinetics Toxicodynamics
Human variability
orPathway-relatedUncertainty factorsor
orprocess relatedUncertainty factorsoror
general default(3.2)
or
general default(3.2)
Pathway-related UFs for main routes of metabolism in humans –intermediate option between default factor and chemical specific adjustment factors
Dorne and Renwick, 2005 Toxicol Sci 86, 20-26
Major Routes of chemical metabolism Major Routes of chemical metabolism and and excretionexcretion
Phase I enzymesCytochrome P-450 ADH Esterases
Phase II enzymesConjugation reactionsCytochrome P-450, ADH, Esterases Conjugation reactions
Glucuronidation
% of Pharmaceuticals Metabolized by Individual Cytochrome P450’s in man
P4502D6 P4501A2P4502A6
P4502C9
Sulphation
N-acetylation (Polymorphic)P4502A6
P4502C19
P4502E1
Amino acid conjugation
P4503ARenal excretion
TransportersCYP2C9 CYP2C19 CYP2D6* Polymorphic TransportersCYP2C9, CYP2C19, CYP2D6 Polymorphic (Extensive and Poor metabolisers, EMs and PMs). *Caucasian 8% PMs 92% EMs
Ratio of internal dose (clearances) between EMs and PMs forCYP2D6 SubstratesCYP2D6 Substrates
Exponential relationships between ratio EM/PM and % CYP2D6 metabolism.PMs covered by pathway-related UFs for substrates with up to 25% (dose) ofCYP2D6 metabolism in EMs
Ratio of internal dose (clearances) between EMs and PMs for CYP2C19substrates
phenotypesphenotypes
70.0
80.0
90.0substrates
40.0
50.0
60.0
EM/P
M
20.0
30.0Ratio
0.0
10.0
0 20 40 60 80 100
% CYP2C19 in EM
PMs covered by UFs for substrates with up to 20-25% (dose) of CYP2C19 metabolism in EMs.
Predicting human variability in toxicokinetics using Monte Carlo modellingMonte Carlo modelling
21
Latin hypercube sampling: variant of Monte Carlo
Stratified sampling throughout the distribution.
Compounds handled by multiple pathways :
1. predict variability and uncertainty factors for healthy adults andsubgroups.
2. Combine distributions describing pathway –related variabilityg p y yand quantitative metabolism data.
3. Compare simulated and published data
Dealing with subgroups subgroupsDealing with subgroups subgroups
-Ratio of internal dose between healthy adults and subgroups
-Pathway-specific variability (GSD).
-Simulate to get the final distributions
Polymorphic pathways : Combine distribution for EM and PM using frequency ofEM and PMs ( for CYP2D6 7 4% PM in Caucasian)EM and PMs ( for CYP2D6 7.4% PM in Caucasian)
PM
combinedEM PM
EMs
Healthy adults:
Uncertainty factors (99th til )Uncertainty factors (99th centiles)
Published Simulated
3 .43 .5
Published Simulated
2 .7 2 .72 .9
3 .03 .0ant ip yrine
co d e ine
d iazep am2 .32 .3
2 .01 .9
2 .02 .1
1 .8
d iazep am
imip ramine
p a race t amo l
p ro g uanil
p ro p rano lo l
Phenotyped healthy adults:
U t i t f tUncertainty factors (99th centile)
3.62.8
3.62.8
2.11 8
2.11.8
codeine propranolol
1.8
codeine propranolol
CYP2D6 EMs CYP2D6 PMs5.2
1.91.8
4.3 codeine
propranolol
Combined EMs and PMs
ToxiokineticsToxiokinetics of binary mixtures: of binary mixtures: CYP2D6 CYP2D6 inhibitioninhibition
20
25
tile)
EM non competitive
PM non competitive
EM Competitive
15
20to
rs (9
5 th
cen
t
5
10
cert
aint
y Fa
ct
0
5
Un
I i t l d i EMIncrease internal dose in EMs.UF for TK (3.2) would not cover EMs for potent CYP2D6 inhibitors.PMs not affected: alternative pathways of metabolism
EMs at risk if metabolite produced the toxicant but reverse situation withinhibition. PMs at risk if the parent compound is the toxicant.
Dorne and Papadopoulos, 2008
HarmonisationHarmonisation of human and of human and Ecological Ecological risk assessmentrisk assessment
Both use uncertainty factors but differ in what/who they aim to protect:Ecosystem or human populations. y p pHarmonisation focus on Mechanistic descriptors e.g., substance parameters, toxicokinetics, toxicodynamics, mode of actionCase studies looking at interspecies differences (mammals, birds) in kinetics
HumanHumanEcologicalEcological
Dorne, Ragas and Lokke. Toxicology 2006
Mechanistic model for ecological species
PORE BODY SOIL
INSIDE ORGANISMOUTSIDE ORGANISMDETOX TARGET
WATER WALL
M
SOIL
MM
M
M M
MMM
MMM
M
M
M M MM
MM
ML
M
MM
MMM
M
ENVIRONMENTAL AVAILABILITY
TOXICOKINETICS TOXICODYNAMICSSpurgeon et al., 2010- STOTEN
Systematic review, meta-analysis
and chemical risk assessment
29
What is a systematic review (SR) ?
• SRs are reviews that attempt to…id tif ll l t t di fitti d fi d it i– identify all relevant studies fitting predefined criteria
– systematically summarize the validity and findings of the studies
– synthesize or integrate the findings
• ...using techniques aimed at minimizing biasg q g
• Governed by principles ofSystematic reviews
y p p– methodological rigour– transparency
Meta analyses– reproducibility
Meta-analyses
30
Questions suited to SR
Type of question Examples of what the question seeks to assess
Effect of a deliberate intervention
- Nutritional properties of an additive in a food or feed
- Efficacy of a vaccine in preventing a diseaseEffect of exposure to a potential risk factor
- Mutagenic effect of a chemical on cells used in mutagenicity tests
A t f Ch i t i ki ti t f tiAssessment of a dose-dependent fate of a substance or dose response
- Changes in toxicokinetic parameters as a function of the dose of a chemical in animals or humans
- Changes in physiological parameters or bi k f ti f th d fdose-response
relationshipbiomarkers as a function of the dose of a chemical in animals or humans (toxicodynamics)
Environmental fate - Changes in the environmental distribution, d d i l hi ff f bdegradation, leaching, or run-off of a substance into surrounding areas as a function of its concentration
Population exposure control outcome (PECO) and steps of risk assessment
• Adult humans (16 or older) P• Chemical in food
PE
• Exposed group, non-exposed group
T i it / id i l f i diti i t d ith d
C
O • Toxicity/epidemiology of a given condition associated with dose; cancer, target organ damage
O
Hazard identification/characterisation• Hazard identification/characterisationTK: fate of chemical in populationTD: toxicity; genotox non-genotox, dose response
32
• Exposure assessment : in some cases SR from literature• Risk characterisation: SR not relevant
SR and hazard identification
Specific questions Question type, open/closed question, key-elements
Answer question using the SR method or SR h? T f id i d?elements search? Type of evidence required?
Does chemical X havegenotoxiceffects/cancer in rat
Narrow, CLOSED question, same question typeas above. Key-elements: chemical X=exposure,human liver=population, genotoxic effect/cancer
Potentially SR. Cohort studies in humans. If notavailable, case control studies. If not aggregateddata (clinical reports) may be considered. If not, dataeffects/cancer in rat
liver?human liver population, genotoxic effect/cancerinduction [=outcome] and comparator [=non-exposure])
data (clinical reports) may be considered. If not, datafor structurally-related compounds.
33Must be determined if SR worthwhile
SR and hazard characterisation
Specific questions Question type, open/closed question, key-elements
Answer question using the SR method or SR search? Type of evidence required?
Dose-response relationshipbetween chemical X and livertoxicity in the rat?
Narrow, CLOSED question, Dose-response type. Key elements:population=rat, measurement 1(quantitative)=dose of chemical X,
Potentially SR . Randomised control in vivo studies in rat using multiple doses over time following GLPs (OECD guidelines). If not available, randomised
outcome=liver toxicityg ) ,control in vivo studies. When none of the above are available, randomised control in vivo studies on structurally-related compounds may be considered.
Dose-response relationshipbetween chemical X and livertoxicity in humans?
Same as above. Key elements:population=humans, measurement 1(quantitative)=dose of chemical X,outcome=liver toxicity
Potentially SR .Ideally, randomisedcontrol trials in humans using multipledoses over time. If not available,aggregated data, clinical reports, narrowdose studies. When none of the above
il bl d t f t t ll l t dare available, data for structurally-relatedcompounds may be considered.
Previous ADI/TDI been derivedfor chemical X?
Narrow, OPEN question This Q is answerable doing a broadliterature search and anarrative description of the results.
34Must be determined if SR worthwhile
SR and exposure assessment and Risk characterisation
Specific questions Question type, open/closed question, key-elements
Answer question using the SR method or SR search? Type of evidence required?yp q
Exposure assessmentHow much ofchemical X occurs inthe different food
Narrow, CLOSED question, Occurence type.Key-elements: quantity of interest=quantity ofchemical x population= food commodity
Occurence data would be required and a SRwould not be necessary. In case not available, thisQ is potentially answerable using the SR methodthe different food
commodities?chemical x, population= food commodity Q is potentially answerable using the SR method.
In case, aggregated data on the concentration ofchemical X could be used
How much of thefood commodity is
Narrow, CLOSED question, quantity ofinterest=quantity of food commodity
Food consumption data over time would berequired and a SR would not be necessary. Iny
consumed byhumans?
q y yconsumed, population=humans
q ycase not available, this Q is potentially answerableusing the SR method. In case, aggregated dataon food consumption could be used.
Risk characterisation
What is the riskassociated withhuman exposure tochemical X?
Complex, OPEN question Answerable doing a broad literature search and anarrative description of the results
35Must be determined if SR worthwhile
SR and Meta-analysis of human data :cadmiumhuman data :cadmium
CADMIUM
Urinary cadmium
reflects thisdose
CADMIUMin food
Accumulatesover years
reflects thisaccumulation
biomarker
kidney Kidneydamages:β2-
effectbiomarker
• SR studies linking internal dose (urinary β2-microglobulin
SR studies linking internal dose (urinary cadmium) to (early) biomarkers of bone/renal effects
• Extensive literature search (19661966 OctoberOctober• Extensive literature search (19661966--October October 20082008) (2 persons in parallel / cross checking)
• Geometric means and SD recorded
36
• 5000 abstracts > 200 relevant papers > 63 included
Final databaseRenal
BiomarkerTotal β2-
MGα1-MG
NAG (total)
NAG a
NAG b
RBP Protein-Uria
(total)(total)
N studies 54 35 16 27 1 2 10 11Continous
data
Bone Total BMD Calcium bALP PTHBoneBiomarker
Total BMD Calcium serum
bALP PTH
N studies 9 5 5 5 4
165 entries
30,000 individualsN studies
Continous data9 5 5 5 4 individuals
37=> 1 to 10 « entries » by study
Hill dose-effect model
Log b2GM
ude Shape
Parameter( )
amplit (η)
background
dLog UCd
ed50
38Effect=bkground + amplitude*(dη / (dη + ed50
η) )
Effect (B2MG) vs dose (U-Cd) data
1000000
100000
/g c
rea)
1 colour = 1 studyDiameters=GSD
1000
10000
lobu
lin (u
g/
10
100
B2-
Mic
rogl
0.1 1 101
10
Urinary Cadmium (ug/g crea)
39
Urinary Cadmium (ug/g crea)
Points considered in the analysis
• Account for group sample sizesAccount for and quantify inter study variability• Account for and quantify inter-study variability
• Account for and quantify the population variability«surrounding » the dose effect curve => allows for«surrounding » the dose effect curve => allows for BMD evaluation for any cutoff
effect Prevalence Correspondingeffect Prevalence Correspondingto the cut off
Cut off
40dosed1 d2 d3 d4
Overall fit
106
104
105
ug/g
cre
a)
103
104
rogl
obul
in (u
102
B2-
Mic
r
10-1 100 101 102101
Urinary Cadmium (ug/g crea)
41
Model fit with adjustment for ethnicity
106
4
105
g/g
crea
) asiancaucasianjapanese
103
104
glob
ulin
(ug
102
10
B2-
Mic
rog
10-1 100 101 102101
Urinary Cadmium (ug/g crea)
42
Urinary Cadmium (ug/g crea)
Assuming additive effect on the log scale
Bayesian model for metaBayesian model for meta--analysis of analysis of toxicokinetictoxicokinetic datadata
Σstudy Μdrug Σdrug σΜdrug σΣdrug Pr(PM) Population cÜ|ÉÜáΣstudy Μdrug Σdrug σΜdrug σΣdrug Pr(PM) Population cÜ|ÉÜá
μ μd
study drug drug drug drug ( )
PM
parameterscÜ|ÉÜá
Population modelsμ μd
study drug drug drug drug ( )
PM
parameterscÜ|ÉÜá
Population modelsμstudy μdrugσdrug
)( k
modelstudy k
drug j
PM Population models• study mean: lognormal• compound mean: lognormal• compound specific inter
μstudy μdrugσdrug
)( k
modelstudy k
drug j
PM Population models• study mean: lognormal• compound mean: lognormal• compound specific inter
μsubjσsubj
)1(~1
,~)log(
)(2)(2)(
2
)(
)(2)(
)(
)(
log−
−
⎟⎟⎠
⎞⎜⎜⎝
⎛∑
jkjkjk
jk
jkjk
ijk
jk
nn
S
nN
nY
i
χσ
σμ
subject i
• compound-specific inter-subject var: gamma• PM subgroup mean:Mixture of lognormals
μsubjσsubj
)1(~1
,~)log(
)(2)(2)(
2
)(
)(2)(
)(
)(
log−
−
⎟⎟⎠
⎞⎜⎜⎝
⎛∑
jkjkjk
jk
jkjk
ijk
jk
nn
S
nN
nY
i
χσ
σμ
subject i
• compound-specific inter-subject var: gamma• PM subgroup mean:Mixture of lognormals
Data: Σlog(Yi)/n ; S2/(n-1) ; n ; (subgroups id)
Mixture of lognormals
Data: Σlog(Yi)/n ; S2/(n-1) ; n ; (subgroups id)
Mixture of lognormals
g( ) ; ( ) ; ; ( g p )g( ) ; ( ) ; ; ( g p )
Dorne and Amzal, 2008, Toxicology Letters, Eurotox 2008
44
In a nutshell… In a nutshell… CHEMICAL
Toxicological effectsExposure assessment
TICA
L ST
RY Occurrence data Food
Genotoxic Non-genotoxic
ANAL
YTCH
EMIS (Concentration in
food)
Food consumption
Toxicokinetics / Toxicodynamics
Chronic Acute
ANTI
TATI
VEOD
ELLI
NG Benchmark dose/ NOAELProbabilistic or deterministic
approach
QUA
MO
Health-based guidance value
pp
MOE ADI / TDI ARfD
45RISK ASSESSMENT Dorne et al (2009) TrAc: 28,6 , 695
CONCLUSIONSCONCLUSIONSCONCLUSIONSCONCLUSIONS
T d• Towards more :
– Quantitative risk assessmentQuantitative risk assessment– Integration of population variability
• Harmonisation of ecological and human risk assessmentsusing mechanistic descriptors
46• Model-based approaches currently used in regulatory bodies throughout the world
1178620753812 t d t b ht1178620753812_expert_database.htm
Disclaimer
47The opinions and views reflected in this presentation are the authors’ only anddo not necessarily reflect the views of the European Food Safety Authority