Introduction to principles of toxicology and risk

assessment

SEAC Training
Helsinki, 29-30/06/2009

Kimmo Louekari
Unit B4, Evaluation

Disclaimer: This presentation does not represent ECHA’s position on the issues covered nor
is it based on expert consultation within ECHA. The presentation includes an introduction to
some of the basic concepts of toxicology and risk assessment. It is also a preliminary
suggestion of the author on how results of risk assessment process could be used for the
Human Impact Assessment under the restriction processes of REACH.
Introduction

• Paracelsus: “…dose makes a poison”,

– Examples:
• 60 µg/day of cadmium may causes a kidney failure
after long-term exposure,
• “The current recommendations to reduce salt
intake to 5 grams per day (i.e. 5000 000 ug/day)
will lower blood pressure” (Cochrane reviews).
– Dose is usually given as mg or µg per kilogram of
body weight per day
Introduction (2)
• In hazard assessment the aim is
– to describe the intrinsic properties (toxicity) of
chemicals. Understanding the intrinsic hazardous
properties can be based on animal or in vitro tests, on
human data, e.g. clinical cases or epidemiological
studies, and on non-testing methods
– to estimate how much of the chemical is necessary to
produce the toxic response in human (or to obtain
DNEL. i.e. a safe level)
• It is necessary to understand the toxic properties
of a chemical to ensure that it can be used
safely.
Distinction of hazard and risk

• Hazard refers to the intrinsic toxic properties of

the substance/chemical
• Risk refers to the likelihood that there will be
toxic effects in the human population
• Risk can not be assessed without data on the
exposure of the relevant populations

In REACH, toxicity data and exposure data are

combined in Exposure Scenarios
Distinction between risk assessment
(RA) and human impact assessment
(HIA) – some preliminary thoughts (1)
The starting point of RA is that
• There is scientific evidence, often based on animal data,
that a specific effect may occur
The outcome of the RA can be
• A crude estimate of the likely hood or/and severity of the
effect or
• RCR or margin of safety.
e.g. 5 cases of specific cancer may occur in 1 Million exposed
persons (the DMEL, level of exposure and number of exposed
people are considered there)
• Dose response curve with respective NOAEL and
LOAEL
• Qualitative effects sensitisation or skin corrosion may
occur in a specific use scenario
Distinction between RA and HIA (2)

In HIA we ask
• How many people are affected
• How much will the effect (cancer, skin corrosion) reduce
their quality of life (NB. Reversibility of an effect)
• What would be the societal and economic costs of the
overall effect
• How should the health impact be valued

However, note that physical/health impact (RAC) and the

following other, e.g. monetary evaluation (SEAC) should
be distinct.
Example of exposure data - benzene

Exposure scenario Duration and Inhalation Dermal

Freq. of Shift average Shift average
activities [mg/m3] [mg/person/da
y]

3) Production of shift length 84 (1) 420

perfumes, use of (assumed), daily (90th (EASE)
benzene percentile)
26 ppm
4) Distribution of shift length 6.8 4.2
gasoline, marine, road, (assumed), daily (90th (EASE)
rail; 1% benzene percentile)
2.1 ppm
Health effects and RC of benzene

• “…an average exposure of 1 ppm over a

working lifetime of 45 years resulted in
estimations which span several orders of
magnitude. From 1.6 – 3.1 cases per thousand
exposed individuals to a range of 0.02 – 0.036
cases per thousand. (Two meta-analysis based
on epi data)
Toxic effects covered in
regulatory testing and in REACH

• Acute toxicity, irritation/corrosion, sensitisation,

repeated dose toxicity, toxicity to reproduction,
genotoxicity/mutagenicity, carcinogenicity,
• Local/systemic effects (systemic means: distributed by
circulation)
• Acute/chronic effects
– Local and acute: e.g. acids and bases on the skin or
in the eye cause irritation or corrosion;
– Systemic and chronic: e.g. long-term exposure to
benzene increases the cancer risk
Toxicokinetics
• There is no specific information requirement in REACH
• The available information has to be used in the CSA
• Covers absorption (skin, g-i- system, lungs), distribution,
metabolism and excretion (urine, faeces etc.) of
chemicals.
• supports the identification and understanding of
– most relevant route of exposure and
– bioavailability
– detoxification of chemicals, mainly in liver and
– determination of the half-life of chemicals i.e. whether a chemical
should be considered accumulative of not.
How is toxicity identified/tested

• Animal testing, e.g. 90day repeated dose toxicity

study where gross pathology, organs, growth, blood and
urine parameters etc are studies. Organ toxicity and/ or
general systemic toxicity can often be observed
• In vitro testing: for skin and eye irritation/corrosion and
genotoxicity testing, possible for some other endpoint in
future
• Human evidence: comes mainly from clinical reports,
poisoning cases, epidemiological data, or from worker
health surveys
• QSAR, read-across and categories: QSAR
modelling tools better suited for some toxic endpoints
(e.g. mutagenicity) than for the others,
 Animal (in vivo) studies

• Mostly used for effect assessment for human health;

• “More or less” good models for humans;

• Different species and strains;

• EU Annex V Guidance, OECD guidance,…

 In vitro studies
• Results from in vitro test may fully or partly replace an in vivo
test if:
- validated according to internationally agreed validation
principles,
- results adequate for C&L, RA.
• The use of results from in vitro methods that have not yet been
scientifically validated (“suitable studies”) depends on whether
such study indicates certain dangerous property and on the
potential risk identified.

• In vitro studies
• Genotoxicity/Cytotoxicity
• Dermal absorption
• Irritation (skin – eye), phototoxicity
• Cellular metabolism/mechanism of action
• Macromolecular binding
• Sperm motility/Embryo culture
 Human data
• Analytical epidemiological studies
Relationship between human exposure and effects: biological
effects marker, early signs of chronic effects, disease
occurrence or mortality
• case control, cohort, cross sectional studies
• Descriptive epidemiological studies (examine
differences in disease rates in relation to differences in
temporal and environmental conditions )
• Case reports (particular effects after exposure; often
single exposure; misuse of the substance)
• Controlled studies in human volunteers

Ethical
considerations!
Meta-analysis
 (Q)SARs
Results obtained from valid (Q)SARs may indicate the
presence or absence of a certain dangerous property.
Results of (Q)SARs may be used instead of testing when
the conditions in Annex XI (1.3) are met:

- Established scientific validity

- Substance within the applicability domain
- Results adequate for C&L and RA
- Adequate and reliable documentation

Not well developed/validated in relation to mammalian

toxicity (skin sensitization, mutagenicity, skin
irritation/corrosion);

Well developed (Q)SARs for phys-chem properties

relevant for effect assessment for human health
There are many types of
effect data

Read Across
(Q)SARs
In-vitro
Endpoint
Information:
Waiving: Annexes V-IX
Existing
• technical
Last resort information
• exposure

TESTING
 Effect Assessment (1)

• Assessment of the inherent properties – potential to

cause harm
• Objectives:
– Determine Classification and Labelling
– Derive levels of exposure to the substance above
which humans should not be exposed – Derived No-
effect Levels (DNELs)
Effect assessment (2)
Involves
• Selection of the key study, for each endpoint to be
evaluated
• Assessment of the human/biological relevance and
severity of the effects observe e.g.
– reversible/irreversible,
– adaptive or harmful/adverse
– consideration of the toxicokinetics
• Evaluation of the dose-response

Is done by industry, and

checked by the Agency for 5% of the dossiers within the

compliance check
Dose response relationship

• If an experiment is well done, there should be a

relation between dose and the effect/response,
i.e. with increased dose/concentration the effect
increases, i.e. more of the exposure animals
have the effects or the effect becomes more
severe
• If dose response relationship can not be seen, it
may be the it was a chance finding or artifact,
then the study data usually can not be alone
used for risk assessment
Reversibility of the effect

• Normally, reversible effect are regarded as less severe

than irreversible ones.
• In some tests, there is a follow-up period to observe
whether the animal recover in time, e.g.
irritation/corrosion studies and acute toxicity studies (e.g.
7 days).
• Reversibility depends e.g. on the regenerative and repair
capacity of cells and sub-cellular structures; e.g.
epithelial cell of gastro intestinal tract have high
regenerative capacity, whereas nerve cells of adults
have completely lost this capacity.
• Covalent binding of a genotoxic substance with DNA,
cancer and damage of central nervous system are
considered irreversible
DNEL=Derived No Effect Level

• Is calculated by dividing the NOAEL/dose

descriptor by the assessment factor
• For example if NOAEL from a reproduction
toxicity study (reduced fertility/sperm count)
would be 150 mg/kg (of body weight of experimental
animal) and assessment factor would be 1000,
DNEL is 150 µg/kg/b.w (for human).
DNEL (2)
• DNEL is needed fro risk characterization (RC)
• In RC, DNEL is directly compared with the
exposure level of the relevant human population
• For example if workers exposure in cleaning
work is 100 µg/kg (normally derived from
inhalation exposure
• Risk characterization ratio would be 100/150,
which is 0.66, i.e below 1
• Thus, it could be concluded and the use of the
substance in that scenario is safe
Assessment factors in DNEL
Aim to cover
• Interspecies (between animal and human) and
intraspecies (among humans)
• Severity of the effect
• Sensitivity of the human (sub-)populations (e.g.
children, pregnant women)
• Quality of the data
• Extrapolation from shorter test to chronic exposure
A specific DNEL might be named e.g. as worker-DNEL
long-term for dermal route
 Basic assessment factors

Interspecies Intraspecies
variation variation

10 10

kinetics dynamics kinetics dynamics

Implications to the HIA and SEA

• When RCR is close to 1 and especially when it

is above it, there may be health consequences
• Uncertainties in the risk assessment and
especially the size of the sub-population
exposure should then be analysed carefully
• This may have been done by the applicant of
authorisation already or it may require more
close examination by RAC and SEAC
• Combination of 1. CMR effect, 2. high RCR and
3. large size of exposure population (and/or wide
spread uses) would be the ”worst scenario”
NOAEL=No Observed Adverse
Effect Level

• May be obtained from any toxicity study where different

dose levels have been applied
• E.g. Dose levels of 10, 150, 500 mg/kg may have been
used in a repro study
– In case 500 mg/kg caused decreased fertility in the reproduction
toxicity study then 150 mg/kg would be NOAEL
• The effect is supposed to be adverse; e.g. irreversible
organ damage, skin corrosion or cancer are adverse
effects, whereas mild irritation of skin or transient change
of liver enzyme levels are considered less adverse.
NOAEL (2)

• NOAEL normally is not obtained from

irritation/corrosion, mutagenicity or sensitisation
study
• In reproduction toxicity study and repeated dose
toxicity study, when there is an adverse effect it
should always be possible to obtain a NOAEL or
LOAEL
DMEL

• Especially genotoxic carcinogens are

considered to have no effect threshold; thus
NOAEL can not be set.
• This implies that exposure needs to be
minimised leading to very strict controls and
setting of DMEL, Derived Minimal Effect Level
• whereas non-genotoxic carcinogenicity is
thought to be a threshold effect and thus, DNEL
can be obtained
DMEL (2)

• Starting point of DMEL derivation can be e.g.

T25 which is a dose that causes the tumour
incidence in exposure am anal to increase by
25% and compared to the control animals
• To obtain DMEL, T25 is divided by every high
assessment factor
DMEL (3)

• The philosophy behind high assessment factor

is the linear extrapolation and acceptable of
small statistical risk, i.e. if the number of
exposed people is not tens or hundreds of
millions, the risk might be only statistical. i.e.
there would be no cancer cases associated with
the exposure to the chemical
Risk Assessment Process
Data evaluation
Exposure assessment Data set Effects assessment

Operational Toxicity data

conditions TGD

Measurements Assessment
modeling factors
IUCLID5
EUSES,
Exposure ECETOC DNEL
estimate TRA
Risk characterization

RCR
Information requirements (1)

• Depend on the amount of substance

manufactured or imported.
• Information submitted to ECHA in registration
should enable classification and labeling as well
as risk assessment
Information requirements (2)
• For example, at the 1-10 t/a level, data on
irritation/corrosion, sensitisation, mutagenicity (only the
Ames test) and acute oral toxicity
• At higher levels, the information requirement for
repeated dose toxicity and on reproduction toxicity are
set
• At the highest tonnage, i.e. above 1000 t/a, data which
enables full assessment of repeated dose toxicity (up to
chronic toxicity), reproduction toxicity (up to two
generation reproduction toxicity test) and mutagenicity
has to be provided.
An example of qualitative grading of the risk level,
based on R-phrases and level of exposure BS 8800

Level of exposure Hazard categories of the health effects (R-phrases)

based on a
questionnaire
slight harm moderate harm, Irreversible extreme harm
Irritation or other effects are possible, asthma, intoxications,
transient mild skin sensitisation, skin carcinogenesis e.g.
effect, e.g. R20, or eye corrosion, e.g.
R21, R36 R35, R41, R42,
R34, R43, R48
Low Low risk Low risk Medium risk

Moderate Low risk Medium risk High risk

High Medium risk High risk High risk

Restriction process
• When a MS CA justifies the restriction proposal, it may
need to, depending on the specific case
– Proceed from the RA to HIA, in order to point out the public
health consequences of the current non-restricted use(s) and
show the benefits of the restriction
– Revise the toxicological evaluation (e.g. DNEL) or the exposure
assessment to demonstrate the there is an uncontrolled or
unmanaged risk, associated with specific uses/products and
thereby with specific Exposure Scenario(s)
– Provide an estimate of the size of the sub-population exposed to
the substance. For that purpose, information of the market
share/penetration, use pattern may be relevant
•
Uncertainties
• Assessment factors are partly regulatory assumptions
rather that scientific, AFs may be over- or under-
protective
– Especially interspecies differences can be substantial, e.g.
different/lacking mechanisms,
• Only those health effects can be observed for with there
are testing/information requirement; low tonnage testing
is very crude
– There are e.g. immunotoxic and developmental neurotoxic
effects which may go unidentified even at higher tonnage
• Most of in vitro methods and QSAR tools are not yet
robust enough.