ELEMENTS OF
RISK ASSESSMENT
Dr. Bernard Goldstein
Environmental
Protection Agency
Risk Assessment Conference
March 18. 1985
-------
WHY A CONFERENCE?
Inform groups affected by EPA
actions about a critical part of its
decision-making process:
risk assessment
• Define elements
$ Describe uses and imitations
ft Demonstrate performance by EPA
COMMENTS
(Introduction by Dr. Goldstein)
Why a conference?
• The Agency vents to tell group* concerned with
BPA action* about the use of riak assessment
in regulatory deciaion Baking.
Why?
•iak assessment ia increaaingly being uaed in
the regulatory process.
People need to develop an underatanding of
riak aasesaaent to underatand EPA decisions.
Failure to cosvunicate adequately in the past
has led to pr obi eats in public understanding of
EPA'a deciaiona (e.g.. Love Canal, EDB).
Finally, I would like to eaiphaaice we are here
to infona you, not indoctrinate.
-------
WHAT IS RISK?
The likelihood of injury, disease,
or death
WHAT IS
ENVIRONMENTAL RISK?
The likelihood of injury, disease,
or death reselling from human
exposure to a potential
environmental! hazard
CGNMBNT3
There are many types of risk. However, at
thia conference we are concerned with chemical
expoaurea—particularly, riaka that people
confront on a day-to-day baaia.
Riak aaaeaament incorporatea both qualitative
and quantitative aaaeaaaientai the qualitative
aaaeaament of whether the riak will occur and
the quantitative aaaeaaaent of the magnitude
of the riak.
•
The need for riak aaaeaament ia uaually
prompted by the reaulta of reaearch or human
experience which auggeata that adverae effect*
•ay be cauaed by the expoaure.
-------
COMPARATIVE RISKS
OF DEATH
Number of Lifetime
Deaths/Year Risks
• Motor vehicle 46,000
accidents
• Home
accidents
• Lung cancer
deaths in
smokers
25,000
80,000
1/65
1/130
COMMENTS
All human activities carry aoae degree of
risk. There are some activities for vhich the
actual riak can be readily Measured, for exam-
ple, awtor vehicle accidents. These are often
activities for which the riak is very high or
for which the number of people exposed to the
risk is very large (e.g., Motor vehicle acci-
dents), and for which the outcome can be
readily attributed to the exposure.
With most chronic chemical exposures, however,
assessment of the risks is more complex be-
cause the association between cause and effect
ia difficult to identify, particularly if the
risk ia low or the nuaber of people exposed is
-------
COMMENTS
Research include* studies of the toxicology of
the substance in question, caae reports of
effects in husans following exposure, huaan
epidemiologic studies of populations exposed
to the substance, and estimation of human
exposure levels.
Such data, however, do not lend themselves
directly to regulatory decision Making or what
ia known within parts of EPA as risk Manage-
ment, for example, research may indicate that
a substance.causes liver damage in rats at a
certain high doae level. Further analysis is
necessary to make a decision about an allow-
able level of expoaure to the aubatance to
protect the public health. A major component
of this step is known as risk assessment.
Part of the risk assessment process involves
making inferencea from often limited scien-
tific data on the basis of what we currently
know about the underlying biological
processes.
It should be noted that risk assessment is
only one element of the ultimate riak manage-
ment process. In addition to public health
considerations, economic, political, and
social consequences are generally considered.
-------
WHAT ARE THE
COMPONENTS OF
RISK ASSESSMENT?
• Hazard Identification
• Dose-Response Evaluation
• Human Exposure Evaluation
• Risk Characterization
COMMENTS
The EPA'* Method i* based on the process de-
scribed by the National Academy of Science*.
• HsKard identification; The determination of
whether a particular chemical ia causally
linked to particular health effecta.
• Dose-response evaluation; The determination
of the relation between the magnitude of expo-
sure and the probability of occurrence of the
health effecta.
• Human exposure evaluation; The determination
of the extent of human exposure before or
after application of regulatory controls.
• Bisk characteritation; The description of the
nature and often the magnitude of human risk,
including attendant uncertainty.
Before describing the individual steps of risk
assessment, I would like to discuss the biological
basis for performing this process ... (next three
slides)
-------
RISK ASSESSMENT
-------
STRUCTURE OF DNA
Normal TraraafclkM Abnormal Transcr^tfon
DMA Prorate DNA Protafa
T T
C Sarina C
<1 • 1
c
(v f a^Aal A ft ^^L.
)-* («*!>!) -»Qly
A
T
T
C
--• r\ . t «i
etna v»^i HI
A
i T
T Asparagina T
A f
1 A
•
• • •
• • •
• • •
Arginina
1
iml-»QlwekiA
VM » ^r\jiy **• 10
1
Asparagina
1
•
•
•
COMMENTS
DMA contain* the coded information that regu-
late* the growth of all organisms.
Damage to DNA or alteration in the eequence of
bases may affect how growth ia regulated. One
posaible outcome of thia ia cancer.
It ia theoretically poaaible that a a ingle
molecule of a carcinogen could produce an
alteration in the DNA aequence at a critical
•ite, which could aubcequently reault in can-
cer. Mathematical modela have been developed
to take thia poaaibility into consideration.
-------
^Threshold
Non-Threshold
No Elf act Organ Damage
COMMENTS
• Non-Threshold: For effect* that involve
•Iteration of genetic Material (including the
initiation of cancer), a very amall exposure
to a chemical may be aufficient to produce an
alteration in one cell. Thia altered cell may
in turn multiply to font a tumor. Thus for
effeeta auch aa cancer it ia commonly believed
that there ia no threshold or no absolutely
aafe level of exposure.
• Threshold: For most other biological effects
it ia generally believed that a certain mini-
mum exposure ia necesssry before any altera-
tion in function of an organ or organ aystea
is produced. For example, a certain minium
number of eella must be damaged before the
compensatory ability of the organ becomes
overwhelmed, resulting in a measurable effect.
-------
HOW DO WE DEFINE
WHETHER A CHEMICAL
IS A CARCINOGEN?
COMMENTS
• Determining whether a substance is * carcino-
gen is usually not • YES-NO decision. This
detersiinstion is based on the total weight of
evidence.
• Rather, substances csn be placed along a con-
tinuum fro* definitely not carcinogenic to
definitely carcinogenic in humane.
• Por example, sodium chloride (i.e., salt) ie i
substance about which we are aa confident ae
science will allow that it is not a
carcinogen.
• Bensidine, on the other hand, ia known to
cause csncer in humans and in the majority of
animal species in which it haa been tested.
• The great majority of other substances fall
somewhere between these extremes and the cut-
off for classifying a substance as carcino-
genic is subject to debate.
(Here you can discuss: 'How much of a tumor
increase is considered to represent a carcino-
genic response?M)
-------
HAZARD
• Review and analyze toxicity data
• Weigh the evidence that a
substance causes various toxic
effects
• Evaluate whether toxic effects in
one setting will occur in other
settings
COMMENTS
Hazard identification includea assessing
whether the data meet currently accepted
atandarda of deaign, conduct, pathologic
evaluation, and atatiatical analyaia.
Do the experimental result*, taken aa a whole,
aupport the concluaion of causality?
Under what conditiona of exposure, in what
speciea, have toxic effecta been obaerved?
-------
TOXICITY STUDIES
Indicate specific organs or systems
of the body that may be damaged
by a substance
Identify specific diseases that a
substance may produce, such as
cancer or birth defects
Describe conditions of exposure
that may result in these effects
COMMENTS
But stance* may cause a * id. range of toxic
effect*. These- mmj ranee • ro» relatively
•icor effects, such as skin irritation,
through variou» foms of organ damage, to
potentially lethal effecta such as cancer.
Individual cheeicals cae cause various forais
of toxicity. For example, consuaption of
alc:ohol can cause central nervous system de-
pression, delirium tremena (DTs), liver
toxicity, and adverse effects on the develop-
ing fetus.
The purpose of toxicity tests is to determine
what foms of toxic effects a chemical may
cause and under which conditions (such aa
route of administration, dose, duration) they
may occur.
-------
SOURCES OF TOXICITY DATA
Himum Studies
Case reports
EpktemfGlogfc studies
• Geographical
• Temporal
Anbnal Shxfes
Qeneral toricfty studfes
• Actile
• Chronte
Spedabed loxldly studtes
• Teratology
Test TubeStudjes
• Mammalon
tXMMENTS
» Ca*'* reports; tocontrol'ed anecdotal report*
on idverse effect* in individuals who have
been exposed to the subsi an :e.
* Epi'iemiologic *f.udiest Con:rolled atudiea
thai: compare the health «ta:us of a group of
pernoaa who have been exposed to a substance
with that of a comparable nonexpoaed group.
atudiea; In general toxicity atudies,
aniawla are exposed to varying doaea of a
substance, by varying routes, and for varying
duration*. During and following exposure,
anlaala are examined to detect varying acute
or chronic (i.e., lifetime) effecta on a range
of endpointa (e.g., survival, body weight,
effect* on • peel fie organa and ays teas).
Animal toxicity studies are conducted baaed •
primarily upon the long-standing assumption
that effect* in humans can be inferred from
effect* in animals. This an sumption haa been
shorn to be generally correct.
—All the chemical* that have been demon-
strated to be carcinogenic in human*, with
the possible exception of arsenic, are car-
cinogens in some—although not all—experi-
mental animal apeciea.
Some effecta (e.g., teratologic, mutagenic)
require special measurement techniques.
-------
DOSE-RESPONSE EVALUATION
Performed to estimate the incidence
of the adverse effect as a function
of the magnitude of human exposure
to a substance
Response
X
,:«l»:
Dose
CCHMENT8
• Dote- response evaluation it generally per-
fom«d to awasurt the advcra • effect that ia
likely to occur *t the loves expoaure level.
• It generally involves two atcpa:
1. Extrapolation froai high to lov doae (in
animals)
2. Extrapolation from aniaula to huaana
• The firat atep generally involves the use of a
•atheaatical aodel to infer riak at low doae
levels.
• The second atep adjusts for differences, such
aa body aixe and life span, between experi-
mental animals and hunana.
-------
DOSE-RESPONSE CURVE
ervable
Range
Range of
Inference
Dose
Rote:
nOHMBNTS
Thi.e elide shots e hypothe icel done-response
cut ve.
The aolid line indicates the shape of the
doie-response c ;irve in the range of riak
levela that ear be practici lly exanined in
experimental studies.
Explain statistical liatitai ions on the levels
of riak that e*... be studied experimentally.
The dotted linea indicate plausible extensions
of the doae-reaponae curve to the low-dose
region in which humana are actually exposed.
Bach of the dotted lines repreaenta a plaus-
ible Mathematical model.
The Mathematical aodel choaen to infer risk at
low-dose levela should be coapatible with our
underatanding of the awchania* by which the
cheaical exerta ita toxic effect (threshold
veraua non-threahold).
-------
FORMS OF
HUMAN EXPOSURE
Inhalation
digestion
Skin Contact
COMMENTS
<• Expiaure May oc ur by v« -it ua routes.
o A aubatance «ay be:
—•Ingeated in f tad or wa :ei
—Inhaled through the ai :
—-Abeorbed through the a*ie
• A peraon «ay be expoaed to a particular
aubatance by any or all of theae routea,
-------
HUMAN EXPOSURE
EVALUATION
• How may people be exposed?
• Through which routes?
• Who is exposed?
• What is the magnitude, duration,
and timing of the exposure?
COMMENTS
• Piri't, we mutt determine in which Media the
aubiitance may bn present (a r, water, food,
soil ).
• MexC, we determine how pcop e may be exposed
to l:he awdia in which the a tbstance ia present
(inhalation, drinking, eating, akin contact).
a lie then identify population groups who maj be
exposed, including their sixes and character-
istics.
a Next, we determine, by actual Measurement or
by environmental fate modeling, the concentra-
tion of the substance in these media and the
amount of human intake through expoaure to
theae media.
• Obviously, there are many uncertaintiea asso-
ciated with each of these atepa. A common
source of this uncertainty sterna from the
distinction between modeling data and actual
monitoring data.
-------
Identification Data
Dose*Responsf»
Evaluation^ Data
Human Exposure
Evaluation Data
Risk
Characterization
COMMENTS
title characterisation:
* Integrates the data collect«d in the firat
thr.se atepa to character: «« the potential riak
to
o Deacribea the atatiatical and biological un-
certainties in eatiMting the extent of health
effecta.
In this final step, riak mmj be characterised in
several vaya:
• Developswnt of an explicit numerical estimate
of the riak for each population group.
• Eatiauition of the Margin of safety for each
population group (i.e., NOEL * estimated
dose).
• Qualitative deacription of riaka for each
population group.
• BOSM combination of these three approachea and
SOM description of their awrits.
Mote: For all four approachea, it is essential that
the atatiatical and biological uncertaintlea
be deacrlbed.
-------
RISK ASSESSMENT ISSUES
Use of animal data
Hazard
Identification
Negative epidemic-
logical studies
ooFMnrrs
Here are BOM iaport^at iaauet • i«ociate4 with
each (tap of the riak ••r-easaent | ro :e«* that I would
like to reeajihasice . . ,-
-------
ADVANTAGES AND
UNCERTAINTIES
OF ANIMAL
DATA
or DIBITS
Humans ar« not jutt big r« :•. They differ in
• Ice, life span, laetabolian, physiology, and
anatoiy. That, ujing animil data involve*
uncertainties related to t'ieia difference*.
-------
RIS!: ASSESSMENT ISSUES
Hazat d Use of animal data
Identi ication
Negative epidemio-
logicaS studies
Dose-Response Extrapolating from high
Evaluation dose to low dose
Extrapolating from
animals to humans
-------
HIGI [-TO-LOW DOSE PROBLEM
R
e
s
P
o
n
s
e
able
Range
Range of
Inference
Dose
COMMENTS
Cener.illy, the avi liable d< tm involve high
«itpoatire levels it aniaiala (or huauma) and
aeaoc ated high r <*ka.
He mint infer how riek altsra with changea in
espoanre levela. Varioua todtla have been
developed baaed or current th :ory from which
to drnw theae inf« rencea.
-------
RISK ASSESSMENT ISSUES
Hazard
Identification
Dose-Response
Evaluation
Use of animal data
Negative epktemWogteal studfes
Eitranoiating from Hgh
dose to low dose
Extranotetmg from
anfcnate to humans
Human Eiposure Modelng vs. ambienl monitoring
EvaluaUon
vs. biological monttoriny
-------
COMMENTS
Bnpoaurs evaluation can be porf^raed baaed
on actual environmental aoni :oilog data or the
application of aathaMtical nxp>aure aodela.
Both Method* involve uncerta.ntr*
Monitoring data often involvua i liaited
nuaber <>f •••pies, on a ••all portion of an
axpoaed population, and liaicatlona in
aenaiti'rity and accuracy of i«t*oda.
Modeling data mtj be aubject to queatioo due
to uncertainty in the nae of •atheaatical
•odela to predict actual environmental
levels.
-------
RISK ASSESSMENT ISSUES
Hazard
IdentiUcaUon
Do*e -Response
Evaluation
Humcn Exposure
Evaluation
Risk
Characterization
Use of anhnal data
Negative epMeiiJolopjteai studies
Extrapolating from Mg0i
dose lo low dose
Extr apotattig from
animals to humans
Moddhg vs. ambient monftonng
vs. biological monitoring
Ouaitattveor
quantitative
-------
RISK CHARACTERIZATION
32? per 1,000,000 exposed people wl dto
from Ifetkne expos*** to Chemical A.
COMMENTS
Thia slide demonstrates ths tange of waya in
whicti riak of cheaical expoaire m»j be
characterised.
In tie top example, riak ia thwracterised in
quantitative term* with no firther
explanation.
In the aecond example, the riak aaaeaaor
determined that the chemical waa carcinogenic
in two laboratory animal apeciea and, through
the uae of lov-doae extrapolation model* and
human expoaure eatimatea, predicted a range of
poaaible riaka to an expoaed population.
In the final example, riak in laboratory ani-
mala ia characterised in qualitative terms
only; no prediction of the riaka to humana ia
made.
-------
R! >K CHARACTERIZATION
32V par 1,000,000 exposed people vrii dfe
fror i If eftne exposure to Ctiemleal A.
Chemnical A It cardnogetito h rate and mice.
Appication of low-dose eitrapototton
moc^eb and human eiposura esttmatea
that the range off rtaka hi humans
ts 1UO-1tOOO deaths per 1JOOO.OOO persona
exposed.
-------
RISK CHARACTERIZATION
927 per IjOOOjOOO exposed people wl db
from if etime exposure to Chemical A.
Chemical A b carcmogento hi rate and mice.
Apfication ol low-dose extrapolation
models and human exposure estimates
suggests thai the range of risks at humans
b 100-1,000 deaths per 1JQOO.OOO parsons
exposed.
Chemical A b cardnogenta b rats and nice
and II b prudent pubac heaiUi polcy to
assume It b abo carcinogenic In humans.
-------
Range of expert judgment
-------
Range off expert judgment
bb-.}:; ft i
aq
*S*fl
------- |