Evaluation of dose–response relationships

[Guest guest]

no single laboratory animal species can be considered the best for

predicting risk of reproductive toxicity to humans in all situations.

http://www.inchem.org/documents/ehc/ehc/ehc225.htm#6.2.2

6.6 Evaluation of dose–response relationships

The evaluation of dose–response relationships is a critical

component of hazard characterization (OECD, 1989; ECETOC, 1992; US

EPA, 1997a; IPCS, 1999). Evidence for a dose–response relationship

is an important criterion in establishing a toxic reproductive

effect. It includes the evaluation of data from both human and

laboratory animal studies. Because quantitative data on human dose–

response relationships are infrequently available, the dose–response

evaluation is usually based on the assessment of data from tests

performed using laboratory animals. However, if data are available

in humans with a sufficient range of doses, dose–response

relationships in humans can also be evaluated.

The dose–response relationships for individual end-points, as well

as for combinations of end-points, must be examined. Dose–response

evaluations should consider the effects that competing risks between

different end-points may have on outcomes observed at different

exposure levels. For example, an agent might increase abnormal sperm

morphology at a low dose, but might decrease total sperm count and

decrease the relative proportion of abnormal sperm at higher doses.

Similarly, malformation or decreased fetal weight might be observed

at a low dose, but prenatal death and decrease in the proportion of

malformed offspring might occur at a higher dose. Whenever possible,

pharmacokinetic data should be used to determine the effective dose

of the target organ.

When data on several species are available, the species most

relevant to humans is the most appropriate for a dose–response

evaluation. This choice is based on several factors, including

comparable physiological, pharmacological, pharmacokinetic and

pharmacodynamic processes; the adequacy of dosing; the

appropriateness of the route of administration; and the end-points

selected. However, information of this nature is often very limited,

and no single laboratory animal species can be considered the best

for predicting risk of reproductive toxicity to humans in all

situations. In some cases, such as in the assessment of

physiological parameters related to menstrual disorders, higher non-

human primates are generally considered similar to humans. In the

absence of a clearly most relevant species, data from the most

sensitive species (i.e., the species showing a toxic effect at the

lowest administered dose) are used, because, as mentioned above,

humans are assumed to be as sensitive as the most sensitive animal

species (Nisbet & Karch, 1983; Kimmel et al., 1984; Hemminki &

Vineis, 1985; Working, 1988; Newman et al., 1993).

The evaluation of dose–response relationships includes the

identification of effective dose levels as well as doses that are

associated with low or no increased incidence of adverse effects

compared with controls. Many studies identify either the lowest dose

causing an adverse effect (lowest-observed-adverse-effect level, or

LOAEL) or the no-observed-adverse-effect level (NOAEL) (Calabrese &

Baldwin, 1994).

Generally, in studies that do not evaluate reproductive toxicity,

only adult male and non-pregnant females are examined. Therefore, it

is often unknown if pregnant females are particularly sensitive to

an agent. In studies in which reproductive toxicity has been

evaluated, the effective dose range should be identified for both

reproductive and other forms of systemic toxicity and should be

compared with the corresponding values from other adult toxicity

data to determine if the pregnant or lactating female is more

sensitive to an agent.

Studies should also evaluate the route of exposure, timing and

duration of exposure, species specificity of effects and any

pharmacokinetic or other considerations that might influence human

exposure. Information should also be obtained as relevant from the

health-related database.

For the developing organism, which changes rapidly and is vulnerable

at a number of stages, it is assumed that a single exposure at a

critical time in development can produce an adverse developmental

effect (US EPA, 1991). Therefore, with inhalation exposures, the

daily dose is usually not adjusted to a 24-h equivalent with

developmental toxicity unless appropriate pharmacokinetic data are

available. However, for other reproductive effects, daily doses by

inhalation may be adjusted for duration of exposure (US EPA, 1996b).

These differences need to be reviewed to determine the most

appropriate approach.

6.6.1 Quantitative dose–response assessment

Quantitative dose–response assessment involves the determination of

a NOAEL or benchmark dose (BMD) and low-dose estimation or

extrapolation. Usually a non-linear (threshold) dose–response

relationship at low dose levels is assumed unless a specific mode of

action or pharmacodynamic data are available to indicate otherwise

(IPCS, 1986c). If sufficient data on mode of action, underlying

reproductive and developmental processes and

pharmacokinetics/pharmacodynamics are available, a biologically

based approach may be used to predict dose–response relationships at

low exposure levels. At the present time, sufficient information is

rarely available for this approach (see Shuey et al., 1994, for an

example). Thus, a chemical-specific approach is used that

incorporates information on the mode of action of a particular

chemical and its pharmacokinetics. In most cases, however, data are

available only on exposure level and associated adverse outcomes. In

these instances, the dose–response analysis consists of evaluating

the dose–response relationships within the observable range and

determining the NOAEL, LOAEL or BMD, then using this information to

calculate a low level of exposure (guidance or reference level) that

is considered to be without appreciable risk (IPCS, 1994). This is

typically done through the use of uncertainty factors applied to the

NOAEL, LOAEL or BMD, but may also be done by low-dose extrapolation

when data are available to support such an approach.

6.6.2 Determination of the NOAEL, LOAEL, BMD and guidance levels

As mentioned above, the NOAEL is the highest dose at which no

adverse effects are detected, and the LOAEL is the lowest dose at

which an adverse reproductive effect is detected compared with the

appropriate controls. These doses are often identified based on

statistical differences from controls, but can be determined by

examining the trend in response and certain biological

considerations, such as rarity of the effect. Evidence for

biological significance can be strengthened by supporting evidence

such as mode of action or biochemical response at low exposure. The

existence of a NOAEL does not indicate that a threshold or non-

linear dose–response relationship exists below the observable range;

it only defines the highest level of exposure at which no

significant adverse effect is observed under the conditions of the

study.

Mathematical modelling of the dose–response relationship is an

alternative approach to quantify the estimated response within the

experimental range. This approach can be used to determine the BMD

or benchmark concentration (BMC) for inhalation exposure, which can

be used in place of the LOAEL or NOAEL (Crump, 1984). The BMD (used

here for either BMD or BMC) is defined as the lower confidence limit

on a dose that produces a particular level of response (e.g., 1%,

5%, 10%) and has several advantages over the LOAEL or NOAEL (Kimmel

& Gaylor, 1988; Kimmel, 1990; US EPA, 1995; IPCS, 1999). For

example, (1) the BMD approach uses all of the data in fitting a

model instead of only data indicating the LOAEL or NOAEL; (2) by

fitting all of the data, the BMD approach takes into account the

slope of the dose–response curve; (3) the BMD takes into account

variability in the data; and (4) the BMD is not limited to one

experimental dose. Calculation and use of the BMD approach are

described in a US EPA (1995) document. Guidance for application of

BMD in the risk assessment process is currently being developed (US

EPA, 1996c), and software for calculating the BMD is available on

the Internet (http://www.epa.gov/ncea/bmds.htm).

Several approaches to calculating BMDs for prenatal developmental

toxicity data have been evaluated ( et al., 1994a, 1994b;

Faustman et al., 1994; Kavlock et al., 1995). These studies apply

several dose–response models, both generic and developmental

toxicity-specific models, to a large number of standard

developmental toxicity studies with dosing throughout the period of

major organogenesis (or, in some cases, throughout pregnancy). These

studies show that such models can be used successfully with prenatal

developmental toxicity data. In a study of the proportion of

implants or offspring affected per litter, the BMD for a 5% excess

risk above controls corresponded on average to the NOAEL. Variables

such as intralitter correlation and litter size appeared to enhance

the fit of the developmental toxicity-specific models. BMDs from

quantal data (i.e., the number of litters affected) for a 10% excess

risk corresponded on average closely to the NOAEL. Various models

were also applied to fetal weight data, and approaches for

determining BMDs for continuous data were established (Kavlock et

al., 1995). BMDs similar to the NOAEL were also obtained using

several different definitions of difference between experimental and

control values. A workshop was recently held on the criteria for

application of the BMD concept. One of the conclusions reached at

the time of that workshop was that sufficient information was

available to begin using the BMD approach for developmental toxicity

( et al., 1995).

The NOAEL, LOAEL or BMD approach can be used to calculate a guidance

or reference level of exposure below which no adverse effects above

background would be expected. These guidance levels include

reference dose (RfD), acceptable daily intake (ADI) and tolerable

daily intake (TDI).

6.6.3 Low-dose estimation/extrapolation

In quantitative dose–response analysis, a non-linear dose–response

relationship (threshold) is generally assumed for reproductive

toxicity (and for many other health effects), unless the mode of

action of the agent is genotoxic. Because of the threshold

assumption, it is not usually appropriate to use mathematical models

to extrapolate to low doses for reproductive toxicity. Instead, a

guidance value is set based on oral, dermal or inhalation data for

chronic exposure. This approach does not estimate risk at a

particular dose level, so when exposure occurs at levels above the

guidance value, there is no way to estimate risk at that dose level;

this is viewed as a significant disadvantage. Because of the short

duration of most studies of developmental toxicity and the fact that

a single exposure may be sufficient to produce a developmental

effect, a separate guidance value for developmental toxicity (e.g.,

an RfDDT or RfCDT) can be determined. However, this procedure is not

followed when developmental toxicity is the " critical effect, " i.e.,

the effect at the lowest dose level.

The guidance/reference value is derived by dividing the NOAEL or BMD

for the critical effect by uncertainty factors that account for such

things as animal to human extrapolation, variations in sensitivity

within the human population, lack of a NOAEL and various

deficiencies in the database. The uncertainty factor is unique for a

given agent, and considerable scientific judgement is required to

arrive at a satisfactory value for this factor. Where there are

adequate toxicokinetic or toxicodynamic data, default values can be

replaced with compound-specific adjustments. As part of the IPCS

project on the Harmonization of Approaches to the Assessment of Risk

from Exposure to Chemicals, a document has been prepared to provide

guidance on the use of such data to replace default uncertainty

factors (IPCS, 2001b).

Taken from US EPA (1996b).

6.8.2 Risk descriptors

There are a number of ways to describe risks for reproductive

toxicity. The first is related to interindividual variability, which

is the range of variability in response to an agent in a given

population and the potential for existence of highly sensitive or

susceptible groups within that population. A default assumption is

that the most sensitive individual in the population will be no more

than 10-fold more sensitive than the average individual; thus, a

default 10-fold uncertainty factor is often applied in calculating

the RfD or RfC to account for this potential difference. When data

are available on highly sensitive or susceptible subpopulations,

their risk can be characterized separately or by using more accurate

factors to account for the differences. When data are not available

to indicate differential susceptibility among reproductive phases or

between males and females, all stages of reproduction are usually

assumed to be highly sensitive or susceptible. Certain age

subpopulations can sometimes be identified as more sensitive because

of critical periods for exposure — for example, pregnant or

lactating women, infants, children, adolescents or the elderly. In

general, not enough is understood about how to identify sensitive

subpopulations without specific data on each agent, although it is

known that factors such as nutrition, personal habits, quality of

life, pre-existing disease, race, ethnic background or other genetic

factors may predispose some individuals to the reproductive toxicity

of various agents.

The second important descriptor is concerned with population

exposure. For example, what portion of the population exceeds the

RfD/RfC, ADI or other guidance value? In some cases, the focus is on

highly exposed individuals. These are individuals who are more

highly exposed because of occupation, residential location,

behaviour or other factors. For example, children are more likely

than adults to be exposed to agents deposited in dust or soil either

indoors or outside, both because of the time children spend crawling

or playing on the floor or ground and because of the mouthing

behaviour of young children. The inherent sensitivity of children

may also vary with age, so that both sensitivity and exposure must

be considered to characterize their risk. If population data are

absent, various scenarios can be assumed for high-level exposure

using upper-percentile or judgement-based values. This approach must

be used with caution, however, to avoid overestimation of exposure.

The third descriptor that is sometimes used to characterize risk is

the margin of exposure (MOE). The MOE is the ratio of the NOAEL (or

BMD) from the most appropriate or sensitive species to the estimated

level of human exposure from all potential sources. Considerations

for the acceptability of the MOE are similar to those for the

uncertainty factor used to calculate the RfD, RfC or other reference

values from the NOAEL or BMD. The MOE has been calculated from

reproductive toxicity data for several chemicals. Examples include

dinoseb (US EPA, 1986), lithium (, 1995) and boric acid and

borax (, 1997). In the case of dinoseb, the MOEs were very low,

in some cases less than one, indicating toxicity in the animal

studies at levels to which people are exposed. This information on

dinoseb led to an emergency suspension of use of this pesticide in

the USA in 1986 and ultimately led to its removal from the market

(Kimmel & Kimmel, 1994, 1996).

Reproductive risk descriptors are intended to address variability of

risk within the population and the overall adverse impact on the

population. In particular, differences between high-end and central

tendency estimates reflect variability in the population but not the

scientific uncertainty inherent in the risk estimates. There is

uncertainty in all estimates of risk, including reproductive risk.

These uncertainties can result from measurement uncertainties,

modelling uncertainties and assumptions made due to incomplete data.

Risk assessments should address the impact of each of these

uncertainties on confidence in the estimated reproductive risk

values.

Both qualitative and quantitative evaluations of uncertainty provide

useful information in a risk assessment. The techniques of

quantitative uncertainty analysis are evolving rapidly. An approach

was recently proposed for estimating distribution of uncertainty in

non-cancer risk assessments (Baird et al., 1996).

6.9 Summary

This chapter summarizes the risk assessment strategies for

reproductive toxicity, including effects on sexual function and

fertility and on developmental toxicity. Guidance on principles and

specific protocols for risk assessment for reproductive toxicity has

been published elsewhere. It is evident that assumptions must be

made in the risk assessment process because of gaps in knowledge

about underlying biological processes and in extrapolating data from

one species to another. The processes of hazard characterization,

quantitative dose–response analysis and exposure assessment must be

integrated in the final characterization of risk, which must be

evaluated and summarized. The approaches described in this chapter

can be applied to thoroughly evaluate the potential for reproductive

risk as a result of exposure. Although a number of advances have

been made in the approaches for reproductive risk assessment, there

are still many gaps in the knowledge base and a need for research to

fill those gaps.

REFERENCES