SCOPE 49 - Methods to Assess Adverse Effects of Pesticides on Non-target Organisms, Chapter 3, Methods to Assess Toxic Effects of Pesticides in Humans

Pesticides, like most other chemicals, have the potential to produce both acute and chronic injury to health. Generally, acute toxic effects are those caused shortly after a single dose or a few doses;* however, some compounds have been known to cause delayed manifestations of injury after a single large dose. Illustrations of delayed toxicity are neuropathy caused by organophosphate esters, lung damage by paraquat, and sensitization by pyrethroids.

Within this framework, acute and chronic effects can be separated in general terms as presented in Table 3.l. These groups form the basis for the discussion in the following sections on the acute and chronic effects of pesticides on human health.

3.1 ACUTE TOXICITY†

Assessing exposure and dose-response relationships for acute toxicity is generally more easily accomplished than that for chronic toxicity. Because acute exposure is restricted, by definition, to a short time interval, estimates of exposure can often be made more confidently in occupational, non-occupational, and laboratory settings. For example, dermal exposure in the workplace is 20 to 1700 times greater than inhalation exposure (Feldman and Maibach, 1974). Since exposure of the lungs to pesticides almost always results in effects at remote sites rather than in the lungs, this organ is not considered to be a major target for acute pesticide toxicity. Dermal exposure, either with or without skin lesions, resulting in systemic effects is of particular concern for public health.


Acute toxicity	Chronic toxicity	Target organ toxicity

Mortality and morbidity	Carcinogenicity	Lungs
Neurotoxicity	Mutagenicity	Cardiovascular system
Immunotoxicity	Development toxicity	Endocrine system
Dermal toxicity		Haematopoietic system
Reproductive toxicity		Musculo/skeletal system
		Liver
		Kidneys

†This section was prepared by S. Baker, J. Doull, J. Finkelman, and A. Massoud.

The doseresponse characteristics for acute effects, unlike those for most chronic effects, are also definable in terms of severity, potency, and existence of thresholds. Consequently, risk assessment or hazard evaluation procedures used to establish margins of safety for acute effects are relatively straightforward and accepted as standard practices by different national and international advisory bodies. In the succeeding sections of this chapter, current methods for determining acute effects noted in Table 3.1 are identified and critiqued for their predictive applicability in assessing risk of acute exposure to pesticides. On the whole, the available methods are highly predictive, although some contain shortcomings that lessen their predictive utility for risk assessment.

Acute pesticide poisoning, particularly in developing countries, is frequent, and thus has an elevated importance in public health. Most intoxications occur in workers. WHO estimates that the worldwide incidence of acute pesticide intoxications has doubled in the last decade, from 500 000 cases per year in the 1970s to over 1 000 000 annually in the early 1980s. The case fatality ratio has been estimated to be between 1.5 and 2 per cent. In some countries, the incidence of pesticide poisoning has exceeded the demographic growth and the rates of agricultural production.

Preliminary data from recent studies conducted in the Central American countries by the Pan American Health Organization (PAHO) suggest that the morbidity of acute pesticide poisoning is greater than expected. In 1987 in Costa Rica's Hospital of Gaupiles, 733 of 5000 (i.e., 14.7 per cent) emergency room patients were treated for acute poisoning; numerous fatalities were recorded. An epidemiologic surveillance program was established in five communities of the same Costa Rican area, having a total population of 1505 (including 528 farmers). These communities are located in a banana-producing area of about 490 hectares where, in one year, 286 tons of fungicide and 107 tons of insecticide (mostly organophosphatesOPsuch as aldicarb) were applied. All of the insecticides applied are considered `extremely hazardous' based on the WHO classification of the toxicity of pesticides. In 1987, 63 individuals of these five communities were clinically diagnosed as having been acutely intoxicated, to varying degrees of severity, by pesticides. While there were no fatalities, 58 poisonings were due to occupational exposure, 4 to accidental non-occupational exposure, and 1 to a suicide attempt (rates of poisoning per 1000 are as follows: occupational, 109.8; non-occupational, 2.65).

Pesticides associated with these cases were OP (48 cases), paraquat (4 cases), mixtures (6 cases), and not identified (4 cases). In Honduras in 1986, field tests on 1100 farmers have shown that about 32 per cent of those examined had a decrease in AChE activity of 25 per cent or more.

The magnitude of the public health problem depends on a number of contributing factors. Among them are the type of crop, pesticide regulations, agricultural practices and technology, awareness of the degree of danger, training to minimize exposure, and availability of medical treatment facilities.

Some crops require more pesticide than do others. For example, growth of cotton in Latin America, Egypt, Sudan, and other developing countries consumes about 60 per cent of total insectides used in these regions. Resistance of pests also contributes to the selection of the pesticide used.

The selection of a particular pesticide is also critical. Quite often, this decision is made by individuals with no professional or technical training, which may lead to improper use of pesticides. Cost of a chemical and its effectiveness (number and quantity of applications) are also important elements contributing to the selection of pesticides. In general, the more toxic a pesticide, the lower its cost.

Most developing countries have inadequate pesticide regulations and the inability to enforce the existing ones. Pesticides are openly marketed, and minor verifications are applied. Restricted pesticides are frequently used for purposes other than those intended or approved.

At times, good agricultural practice requires less pesticide usage than is the common practice. Large areas containing single crops such as cotton or bananas are usually treated by professional pesticide applicators. This situation does not imply that all precautions are taken. Quite often, aerial spraying will not spare neighbouring residential areas, Smaller farms do not use professional pesticide applicators. In these cases, the farmer applies the chemicals, which are usually mixtures selected by the individual.

Lack of vital knowledge of toxicity and of procedures for the safe use of pesticides plays a critical role in obtaining direct exposure. Furthermore, labelling practices in developing countries are often inadequate.

Since most of the exposure events occur in distant rural areas, medical facilities to diagnose and treat the patients are likely to be quite distant. In the absence of locally trained personnel, there is greater likelihood that increased mortality results from acute over-exposure.

Typical cases of acute pesticide poisoning occur in males between 14 and 50 years of age, who are usually applying OPs with a backpack sprayer. If they live in an area of Latin America where pesticides are used intensively, their chance of acute intoxication (from clinically detectable to severe intoxication) on a yearly basis is about 10 per cent. This percentage may be as high as 30 per cent, if one includes subclinical levels of intoxication such as AChE depression.

3.1.1.2 Methods for investigation and surveillance of morbidity and mortality

From a practical perspective, the first goal is to reduce to the minimum possible the number of fatal cases caused by pesticides; the second goal is to significantly reduce the risks that may lead to acute intoxication. Bearing these goals in mind, the strategy recommended is to establish proper surveillance programs that take into account the role and needs of all involved parties.

At national and state levels, it is essential firstly to decide which pesticides will be licensed, restricted, or banned, making certain that proper resources are made available to supervise compliance with these decisions; secondly to communicate and enforce sound labour practices regarding the safe use of pesticides; and thirdly to establish reference laboratories to support field testing activities of pesticides.

At the local level, a surveillance program requires additional efforts. In an effective surveillance program, the first step is collection of the following data:

Major difficulties exist in performing epidemiological studies on pesticide exposures, including the assessment of doses, the variability and unpredictability of agricultural exposures, and, in some instances, the migratory lifestyle of some of the agricultural workers. For epidemiological surveillance to be effective, measurements of exposure and characterization of risk must be improved. Design of epidemiological studies depends on the nature of the questions to be addressed and the availability of time and resources. Once the exposed population groups are defined, population registries might be established for follow-up and for cohort studies. These registries might contain information on periodic medical examinations of individuals obtaining high doses. Epidemiological follow-up of individuals exposed or diagnosed as having acute poisoning may stimulate further studies of acute delayed toxicity.

The traditional means of measuring acute toxicity have been the observation of mortality per unit of single dose in groups of experimental animals; the details for such tests have been reviewed extensively by others, and are not addressed herein. Specialized tests for neurotoxicity and immunotoxicity have been developed recently, and will be discussed in some detail below.

Pesticides may produce either acute or chronic neurotoxicity in non-target species. A single exposure can produce immediate symptoms of neurologic poisoning, as well as delayed manifestations of injury. Chronic, low-level exposures usually produce symptoms of chronic poisoning and, in some cases, even acute poisoning, because the body burdens of the pesticide are so large.

Most cases of acute poisoning from pesticides in human result from exposure to insecticides rather than herbicides, fungicides, and other pesticide classes. The reason is that the neurotoxicity of insecticides (such as the organochlorine compounds, OPs, carbamates, and pyrethroids) results from the primary mechanism of action being at either or both the peripheral and central nervous systems, leading directly to motor, sensory, autonomic, cognitive, or behavioural disturbances. By contrast, neurotoxicity from other classes of pesticides is usually a secondary, non-target effect.

Neurotoxicity testing is aimed at detecting and quantifying changes in both structure and function of the central and peripheral nervous systems that are produced by exposure to a test chemical such as a pesticide. Although there are many animal testing methods for detecting neurotoxicity, few methods have been sufficiently standardized and validated to serve as protocols for regulatory purposes. Most current neurotoxicity test methods use indices of neurochemistry, neurophysiology, neuropathology, behaviour, and specific psychological effects (such as perception, motivation, habituation, learning, and memory) to detect toxicity. To adequately assess neurotoxicity as an adverse non-target effect of pesticides, simple non-invasive methods are needed to identify all types of neuronal dysfunction and to quantify effects such as impaired vigilance, reduced concentration, memory deficit, linguistic disturbances, depression, and irritability.

Depression of AChE activity in serum, whole blood, or erythrocytes has been used extensively to detect exposure to cholinergic pesticides (i.e, carbamates and OPs) and to quantify the magnitude of exposures. Interpretation of these findings requires some clinical judgement, since the effects depend on the class of pesticide (e.g., carbamates produce a briefer depression than do OPs), the specific agent (some OPs depress serum esterase but not erythrocyte esterase); the route and duration of exposure (repeated and prolonged exposure at relatively low doses may depress esterase without producing cholinergic symptoms); the effect of treatment (oxime reactivators such as 2-PAM will modify the esterase effect); and the formulation (concomitant exposure to mixtures of some pesticides may produce either potentiation or inhibition of the effect). Nevertheless, measurement of AChE inhibition is a useful and well-validated method for assessing exposure; it can be used in the field for relatively low cost.

Delayed neurologic sequelae (e.g., paraesthesia, tremors, and equilibrium disturbances) have been reported with organochlorine pesticide exposure; delayed neuromuscular paralysis which can be detected by non-invasive nerve conduction studies has been produced by exposures to OPs such as phosvel, leptophos, and mipafox. In vivo tests, such as the leghorn chicken test, and in vitro enzyme studies can be used to detect the neuromuscular effects of OPs, and are useful to predict similar effects in humans. However, other neuronal dysfunctions may also occur as a delayed response to acute or chronic exposure to cholinergic insecticides.

Acute exposure to the organochlorine insecticides such as DDT, cyclodienes, and chlordecone may produce acute convulsive seizures with residual neuronal cell loss and brain ischemia and anoxia. These effects can result in a variety of behavioural changes (irritability, loss of recent memory, depression, anxiety, EEG modifications, and learning impairment in humans; difficulty in performing complex tests in animals), which may persist for many years. Chronic exposure to these agents can produce similar effects, but the persistence and severity of these changes appear to be related to the incidence of convulsive seizures in both acute and chronic exposure situations.

Aside from producing hypersensitivity, relatively little indication exists that pesticides produce anything other than minimal dysfunction of the immune response in humans. Current technologies make it possible to detect immunomodulation or disease that results from modulation of the immune system.

Pesticides were among the first environmental chemicals to be tested for immunotoxic effects. Because the immune systems is so complex, a single test is inadequate to evaluate immune alterations caused by pesticides. Rather, a battery of tests is required to examine specific endpoints. The occurrence of immunotoxicity is influenced by pre-existing toxicity; conversely, the immune response might predispose individuals to diseases such as cancer. This situation suggests the need for a doseresponse characterization of the relation between pesticides and immune function.

Preliminary screening tests for immunotoxicity before undertaking detailed testing is scientifically reasonable firstly because specific immune function tests are costly and time-consuming, and secondly because relatively little is known about the ability of the majority of pesticides to alter specific immune functions, thus creating a need for broader-scale screening for effects on the immune system.

Advantages have been identified for several screening tests. These tests can be conducted as a part of subchronic or chronic dosing regimens, few experimental animals are required, and a measure of immune response can be obtained by weighing critical organs and performing simple pathological examinations on them. From regulatory and risk assessment perspectives, however, relatively few endpoints included in subchronic toxicity studies are direct indicators of immunotoxicity.

Evidence of immunomodulation is extensive. Specific points along the sequence of events characterizing immune response are susceptible to pesticidal influence. These loci include antigen contact and macrophage processing; antigen recognition by T- or B-cells; clonal proliferation of immunocompetent cells; production, storage, and release of lymphokines; cell-cell communication; and tissue damage. Sensitivity may vary among different cell populations, leading to the possibility of a biphasic immune response (immunostimulation at low dose; immunosuppression at high dose). Such a phenomenon makes tests for immunotoxocity mechanisms highly desirable for understanding the immune response potential of pesticides.

From risk assessment and regulatory perspectives, relatively few immunologic endpoints are considered useful. For a preliminary screen, however, the following are useful: differentiated leucocyte count, total serum protein, organ weights and cellularity, histopathology, and haematology as a measure of pesticide-induced anaemia. Additional tests, such as serum immunoglobulin levels, and possible premature mortality of dosed animals from immunosuppressive effects would yield more definitive screening results.

If the screening procedures show an immunologic alteration, then more comprehensive data to support risk assessment would be required. These tests include host resistance, both specific and non-specific cell-mediated immune response, and time-course for recovery from immunologic effects.

Several clinical endpoints are useful for screening tests. Haematology, spleen cell, and bone marrow evaluations can provide clues to an altered immune response via lymphocyte transformation, mixed lymphocyte response, or in vitro antibody formation. In vitro macrophage functions can also be evaluated in pesticide-dosed animals. While in vitro phagocytosis and immunoglobulin profiles (identified with ELISA) are also useful, these techniques are less sensitive for evaluating immune response after antigenic challenge, because they are not as responsive as the immune system sensitized by pesticide pretreatment. Finally, host-resistance models, including the use of tumour cells, bacteria, viruses, blood parasites, or E. coli endotoxin, may indicate a thymus-dependent immunity and thus may also be useful as a screening tool.

Batteries of targeted tests are available for comprehensive investigation of immunological effects. These include tests for humeral and cellular immunities, macrophage responses, and biochemical mechanisms. In each intance, the following features must be carefully crafted or considered for proper application of a test:

Techniques involving the use of human cells and in vitro metabolism are somewhat unreliable, because of inconsistency due to individual variation and potential confounding effects.

Autoimmunity has been reported in workers with chronic pesticide exposure. The development of autoantibodies against liver and kidney occurs after organochlorine or OP insecticide exposure and usually follows pathologic changes in tissue.

Similarly, hypersensitivity in humans has been reported. However, in contrast to autoimmunity, for which no reliable animal model has been identified for testing, there is an experimental basis for examining hypersensitivity in animals. Cutaneous or pulmonary hypersensitivity in animals caused either by pesticides or industrial solvents has been observed. Standardized protocols for detecting hypersensitivity in the guinea pig are available from USFDA and OECD.

The skin is one of the main routes of systemic exposure to pesticides. Feldman and Maibach (1974) showed that spraying or dusting of pesticides causes doses 20-1700 times greater than that reaching the respiratory tract. The lipophilic component of the skin diffuses non-polar molecules, while the intracellular proteinaceous material of the skin permits diffusion of polar molecules, a process that is enhanced by the moisture content of the outer layers of the skin.

Many factors play a role in the percutaneous absorption of pesticides: among them are the nature of a chemical substance (e.g., chemical structure, concentration, total dose, application frequency, and duration of contact), and the characteristics of exposed individuals (e.g., individual variation, temperature, humidity, surface area potentially exposed, site of application, and skin condition including pH and cleanliness). Dermal absorption of pesticides can be greatly enhanced by solvents or other ingredients in a formulation.

In the clinical setting, cases of acute exposure may show different symptoms. Contact dermatitis (Baginsky, 1978) is the most common of these symptoms, manifested either as non-allergic irritation or allergic contact dermatitis. This latter effect is a form of delayed hypersensitivity resulting from the low molecular weight of a chemical, usually a hapten conjugated with protein in blood (Morison et al., 1981). Localized or generalized urticaria, pigmentation (discoloration, increase or decrease of melanization), and occupational acne (halogen acne and chloracne) are also known to occur (Yonemoto et al., 1983; Echobichon et al., 1977). Cutaneous parasthesia (i.e., delayed transient itching sensation mostly on facial skin) has been reported in case of acute exposure to synthetic pyrethroid (Knox et al., 1984).

The previous discussion documents toxicological and epidemiological evidence of acute effects related to acute exposures to pesticides. For other organs and systems (e.g., pulmonary*, haematopoietic and haematological, cardiovascular, renal, endocrine, and musculoskeletal) evidence of toxicity or its absence is often lacking. While the absence of data may indicate that these organs or systems are unaffected as a result of a single dose of diverse chemical entities, it is possible that the available methods may be insufficiently sensitive or specific to detect subtle changes in these organs. Thus, before concluding that acute toxic effects are absent in these systems, research should be undertaken to remove uncertainties that surround the issue.

A promising approach to demonstrate subtle effects in these organs and tissues is the use of biomarkers. These techniques have the advantage of providing non-invasive, inexpensive, field-adaptable exposure assessments that could be particularly useful for risk assessment and epidemiological applications. Such procedures also should be useful to identify susceptible populations such as the young, the elderly, or those with pre-existing conditions or predisposition to disease.

Toxicokinetics (the dynamics of absorption, biotransformation, and excretion) represents another promising approach to detect the adverse effects of both acute and chronic exposures. Toxicokinetic information is likely to have particular applicability to single-dose exposures where factors, such as an enzymatic pathway of detoxification, are rate limiting (i.e., saturation kinetics). For example, failure of the kidney to demonstrate an adverse response to acute pesticide exposure may reflect the absence of saturation kinetics which enhances its capacity and efficiency of elimination. By contrast, other primary pathways of exposure may be saturated, leading to the manifestation of secondary toxic responses that would not occur usually, were it not for saturation of the primary target organ or pathway. For example, saturation by one compound of either the sulfhydryl deactivation process or the P₄₅₀ microsomal detoxification mechanism may lead to adverse effects by a second parent compound. Better understanding of these processes would help resolve some of the more difficult scientific issues important for regulation (such as the establishment of a maximum tolerated dose (MTD) for a compound with multiple, dose-dependent metabolic pathways) or for the appropriate design of animals studies. Ideally, an understanding of the mechanisms of toxic action of pesticides and of their specific relation to the consequent adverse effects remains the most effective tool for making interspecies comparisons, conducting low-dose extrapolations, and increasing the predictive value of the risk assessment process.

3.2 CHRONIC TOXICITY*

This section deals with the chronic deleterious effects of pesticide exposure. These effects may involve the delayed manifestations of short-term, high-dose exposure or be the outcome of continuous low-level exposure. For some compounds, both acute and chronic exposure involve the same target organ being affected predominantly and similar pathological lesions; however, the toxic potency is often quite different. For instance, short-term, high-level exposure to a carcinogen is generally far less effective at inducing tumours than continuous, low-level exposure for the same total lifetime doses.

This section focuses on four specialized areas in chronic toxicity: reproductive and developmental toxicity, genotoxicity, cancer, and target organ toxicity. Several factors are common to each area:

* This section was prepared by R. Albert, R. Kroes, A. Lichacev, M. Mehlman, B. Ordonez, and B. Schwetz.

Knowledge of biotransformation and toxicokinetics of a chemical is vital, since it provides unique insight into mechanism(s) of action of a chemical and its metabolism in target organs. Toxicokinetics also provide a necessary understanding of the relationship between dose and toxicity. Accumulation in the body of pesticides and other chemicals greatly influences the ability of a substance to cause chronic injury; therefore, the extent of accumulation should be determined early in toxicity testing. The degree of tissue accumulation and the steady-state levels of lipophilic pesticides not biotransformed to hydrophobic metabolites define in part the duration of dosing in chronic toxicity tests.

Physiologically-based pharmacokinetic modelling is useful for risk assessments because it describes in quantitative terms the emporal behaviour of a chemical in various body compartments. Analysis of target sites for a parent compound and its metabolite(s) provides information with which to facilitate extrapolation among species, relying on empirically-derived correlations in various species.

The most frequently observed manifestations of reproductive toxicity caused by chemicals are abortions in females and infertility in males. The apparent frequency of these manifestations is due largely to the greater ease of detecting these adverse effects compared to other forms of toxicity. Manifestations of injury to embryonal and fetal development include malformations, decreased birth weight, mortality of the fetus or neonate, and functional changes observed in neonates. The nature of an adverse effect on reproduction and development depends on the chemical, amount and duration of exposure, age, stage of pregnancy, and pre-existing health status. Acute exposure is sufficient to cause significant risk for developmental toxicity because of the narrow `time windows' of susceptibility. In contrast, repeated and prolonged dosing is more critical for risk of reproductive injury.

To date, most changes in human reproduction or development in humans have been detected initially via anecdotal information or case reports. For example, astute clinicians were largely responsible for the early detection of several noteworthy human teratogensthalidomide, alcoholic beverages, diethylstilbestrol, and methyl mercury. The effect of dibromochlorpropane (DBCP) on male fertility was first recognized in humans by chemical workers sharing observations about their inability to have children. Many chemicals which are now recognized as human developmental toxicants were first identified as such in animal studies, and were subsequently confirmed in humans by case reports.

Surveillance records and birth defects registries hold promise for future identification of reproductive and developmental toxicants, but to be useful the databases must be much larger than they are today, and researchers must ask the correct questions to produce meaningful answers. Epidemiologic studies have been helpful to confirm chemicals as human developmental toxicants, but have had limited usefulness in their initial identification (Schardein et al., 1986).

Adverse effects of chemicals on reproductive organs and function are detected by a variety of test methods. The most direct measurement is made from fertility trials that are part of reproduction studies. Single or multiple generation reproduction studies, having typically one or two sets of litters per generation, are most often conducted in mice and rats.

A protocol for reproductive assessment by continuous breeding of rats or mice, recently developed and validated by the National Toxicology Program in the United States, is another definitive test for reproductive toxicity. According to this protocol, mating pairs of rodents are housed together continuously for 100 days; observations include mean number of litters per mating pair, litter size, body weight, gross appearance, and time between litters. If an effect is noted in any evaluated endpoint, animals are cross-mated to determine whether the effect is transmitted by the male, the female, or both. In this continuous breeding study, rats and mice normally deliver between four and five litters during the period of cohabitation. The last set of litters is saved, raised to sexual maturity, and mated to determine fertility of the F₁ offspring.

A large decrease in sperm count can be tolerated by rodents without altering fertility, whereas human fertility is known to be affected by smaller changes in sperm count. Thus, the absence of an effect in rodent fertility does not preclude the possibility of an adverse effect of importance in humans, even though environmental levels of chemicals are typically much lower than the dose levels used in toxicity studies. The sensitivity of rodent reproduction studies is increased through the use of additional measurements on sperm and testis. Changes in testis weight, sperm motility, and sperm count have been found to be correlated positively with changes in fertility of rodents (Morrissey et al., 1988). Significant changes in these parameters, even in the absence of an adverse effect on rodent fertility, should be considered predictive of a potential effect on human fertility. Compared to the database for males, our knowledge of the effects of chemicals on reproductive mechanisms in female humans or animals is quite limited. Improved knowledge of normal performance is needed along with enhanced methods to detect adverse effects on reproductive organs and performance in females.

Studies other than those described above also provide useful reproductive toxicity data. Subchronic toxicity studies routinely include histopathologic examination of reproductive organs and measurement of testis weight. Dominant lethal studies, while designed to detect chromosomal damage in males which results in death of the conceptus of untreated females, also provide useful information on male fertility.

In addition to measurements of effects of chemicals on reproductive organs, function, and performance, another adverse effect of considerable consequence to reproduction is the induction of germ cell mutations. While mutations in somatic cells might lead to cancer, those in germ cells may lead to heritable changes of great pathological significance to the offspring. Studies such as the heritable translocation test, the specific locus test, and the dominant lethal test are the most definitive for germ cell mutations. A number of potent mutagens identified by short-term mutagenicity screens have also been shown to cause germ cell mutations.

The traditional measures of developmental toxicity in animals and humans include malformations, decreased body weight at birth, death of the embryo, fetus or neonate, and alterations of function. Many chemicals shown initially to be developmental toxicants or teratogens (i.e. inducing major structural malformations) in laboratory animals have later been demonstrated to cause the same effects in humans; many agents known to be developmental toxicants in humans act similarly in laboratory animals (with the exception of coumarin derivatives). Thus, chemicals which adversely affect the developing embryo or fetus of animals should be considered potentially toxic to developing humans.

The traditional screening test for developmental toxicity is designed similar to the protocol for the segment II study of the US Food and Drug Administration, a protocol which has changed only minimally over the past 25 years. Studies are conducted in rats, mice, rabbits or hamsters; exposure is generally during embryonal and early fetal development. In all species, fetuses are removed just prior to birth for examinations for external, visceral, and skeletal defects. These studies measure primarily effects on fetal structure, weight, and survival.

Supplemental studies, such as behavioural teratology studies, must be conducted to measure functional changes. In this type of protocol, animals are exposed during organogenesis; the offspring are permitted to be delivered, and are examined for developmental landmarks (tooth eruption, eye opening, vaginal opening, hair growth) and for specific behavioural attributes at various stages during their maturation (auditory and visual function, motor activity, learning skills). Protocols for other functional endpoints are not as well developed as for behavioural teratology.

Short-term in vitro tests for developmental toxicity are not used as routinely as genetic toxicity screens to predict carcinogenicity. Systems using cultured whole embryos or limb buds, insects, hydra, xenopus larvae, or cultured cells or organs have been evaluated; many of these systems, however, receive more use in mechanistic research than as screens or pre-screens for developmental toxicity. A short-term in vitro test first described by Chernoff and Kavlock (1982) and variations of this test are useful to anticipate the needs for reproductive and developmental toxicology tests (Hardin et al., 1987).

While developmental toxicants in humans produce similar responses in laboratory animals (with the exception noted above), animal studies are relatively ineffective at predicting the same type of response in humans (Schardein et al., 1986). For example, from the chemical induction of cleft palate in mice, one cannot infer that the same compound will produce cleft palate in humans. A significant effect in any of the four major manifestations of developmental toxicity is considered sufficient to predict some potential adverse effect in humans. The dose level at which developmental toxicity occurs relative to the dose level at which maternal toxicity occurs is a critical determinant of the type of toxic response seen in another species; there is greater risk to humans from chemicals capable of causing developmental toxicity in the absence of maternal toxicity than from chemicals which are embryotoxic solely as a consequence of maternal toxicity. There are at least two reasons for this:

Many mechanisms exist by which chemicals can cause birth defects, but the critical insult which accounts for the adverse effect on the embryo is known for only a few chemicals. Mutagens are considered to account for very little developmental injury observed in humans (Schardein, 1985).

In contrast to carcinogenesis, developmental toxicity is considered for several reasons to be a process for which there is a threshold. Mutations are not necessary to cause developmental toxicity; most abnormalities are the result of interference with several critical events. Also, the plasticity of the developing embryo makes it very unlikely that a single event in a single cell would result in a recognizable birth defect. As a consequence, extrapolation from animal studies to humans has traditionally been based on some margin-of-safety (MOS) applied to the no-observed-adverse-effect level (NOAEL) in the most sensitive species of laboratory animal tested. A margin-of-safety of 100 or more is usually used, the actual value being selected on the strength of the scientific evidence and the nature of the human exposure to a given chemical.

Compared with that for developmental toxicology, less information is available regarding the agreement between animals and humans for reproductive toxicity. However, limited data are available which support the proposition that animal studies are predictive of at least the types of adverse reproductive effects seen in humans (Amann, 1982; Working, 1988).

The role of route of exposure on the causation of adverse reproductive or developmental toxicity has not been fully investigated; unquestionably these two forms of toxicity are the result of systemic dosing and not of a site-specific effect at a portal of entry into the body (such as may be the case for lung cancer or for dermatitis). Consequently, route of exposure is critical only to the extent that it determines the amount of an agent (parent or metabolite) absorbed into the blood.

Exposures to multiple developmental toxicants is as much a cause for concern as exposures to multiple toxicants with different target organs. The combined toxicity of most mixtures is described by additivity models. When interactive toxicity occurs, either potentiation or antagonism may result, and each is important for public health. Fortunately, each is the exception rather than the rule.

Historically, in vivo laboratory models were developed to detect alterations of transmitted genetic effects. Recently these tests have been directed at, and used as tools to predict, possible carcinogenicity in laboratory animals. By contrast, in vitro, short-term tests were further developed, and used as possible supplements or replacements for the identification of possible carcinogens. The introduction of a rapid bacterial test by Professor Bruce Ames has been the impetus for the investigation of possible genetic interactions that might be predictive of carcinogenicity. In a decade, important developments led to plant, microbial, or mammalian cells being used in numerous in vitro tests. Other short-term tests such as those aimed at detecting promoting properties (e.g., cell-to-cell communication and cell conductivity) are not discussed here, because this section is deliberately limited to methods to detect genotoxic activity in vitro and in vivo.

A drawback of in vitro tests is their inability to metabolize chemicals whose activity is dependent on initial biotransformation to active electrophilic metabolites that react directly with genes. This limitation was overcome partly by introducing subcellular fractions of liver homogenates to mimic effectively in vivo metabolism. Many chemicals require metabolic transformation to produce toxicity in an organism. Indeed, the development of reactive metabolites is believed to be the first step in chemical carcinogenesis, which is believed to be an irreversible alteration of the genome leading eventually to the formation of daughter cells that carry somatic mutations which may later become cancerous.

Numerous plants and organisms have been suggested to detect genetic damage. Bacteria, yeast cells, or mammalian cells in culture have been used far more than have plants or plant cells. The Salmonella systems are the most frequently used as models for prokaryotes; mammalian or yeast cells are most widely representative of eukaryotes, which are closer phylogenetically to humans. Mouse lymphoma cells (L5178Y), hepatocytes in vitro, cultured Syrian hamster embryo (SHE) cells, and Balb/c 3T3 cells are used frequently. Since the cells are either devoid of, or deficient in, metabolic transformation, a liver homogenate must be added to provide that dimension. For most chemicals needing metabolic activation, the inclusion of metabolizing enzyme systems in these assays has been necessary to detect genetic damage. While the inclusion of a metabolic activation system has proven successful, such a system does not completely replace transformation in vivo.

In the in vitro assays a variety of endpoints can be examined, such as reverse and forward mutations, frameshifts and deletions, chromosome breaks and transpositions, sister chromatid exchanges, unscheduled DNA synthesis, and neoplastic transformations.

Since uncertainty exists concerning the mechanism(s) involved in the induction of neoplastic disease, certainty is low about which genetic toxicity test(s) may be most predictive for carcinogenesis. Therefore, a spectrum of in vitro tests that measure different endpoints of genetic toxicity have been proposed for use. Preferably, such a battery should include measurement of at least gene mutations, chromosomal mutations or damage, and DNA damage.

In the past decade, top priority has been given to validation of short-term tests and to the development of laboratory tools whose results could be reproduced around the world. The value of short-term tests for genetic damage as a prescreen for carcinogenicity is now well recognized. When first used, these tests appeared to demonstrate a high degree of concordance between genotoxicity and carcinogenicity. Later evaluations based on collaborative studies indicated far lower concordance. For non-genotoxic carcinogens, short-term tests for genotoxicity will, by definition, be negative. Presently, only chemicals likely to cause cancer by attacking DNA are capable of being identified with some precision using carefully selected short-term tests. On the basis of structureactivity analysis to identify sites on a chemical reactive with genetic material, a correlation (perhaps as high as 60 per cent) is said to exist between the results of the Ames-Salmonella assay and genotoxicity.

As a consequence, when properly validated, in vivo mutagenicity assays may hypothetically replace some lifetime carcinogenicity tests.

In vivo tests for genotoxicity have been developed using insects and mammals, primarily rodents. Tests with insects, primarily with Drosophila, are relatively inexpensive and rapid. Large numbers of test subjects can be used; because of their relatively short lifespan and generation time, results of chronic exposure may be obtained within relatively brief periods of time. A disadvantage of the insect as a test model for humans is the often considerable difference in enzymatic mechanisms that activate and deactivate chemicals. Mice and rats are the mammalian species of choice. When only data from mice or rats are available, extrapolation of results to humans is necessary, despite the existence of numerous uncertainties, some of which are related to differences in toxicokinetics and biotransformation.

Several endpoints can be identified in the systems mentioned above. Whereas in Drosophila sex-linked recessive lethal mutation is the prime target of investigation, chromosomal aberrations and other discrete endpoints can also be studied. In animals, different cytogenetic effects can be studied, such as chromosome aberrations (including appearances of micronuclei), sister chromatid exchanges, and germ cell mutations. In addition, unscheduled DNA synthesis and adduct formation can be studied. DNA adduct formation seems to be a most promising assay for the detection of both genotoxic properties and unknown exposure to genotoxic chemicals in humans.

All genotoxic compounds including some pesticides bind covalently with cellular macromolecules, and form corresponding adducts with nucleic acids or proteins. Depending on the structure of adducts, various methods of monitoring are applicable.

Immunochemical and postlabelling methods are the most relevant to detect adducts formed in vivo by pesticides. Immunochemistry also permits detection of adducts with either nucleic acids or proteins, whereas postlabelling detects adducts with DNA only. Adducts can be identified precisely only by comparing them with corresponding reference compounds; because the number of reference compounds available for postlabelling is limited, the number of adducts that can be identified is also limited.

For each type of method, the identification of adducts of interest is the greatest challenge, since only a fraction of them are believed to be biologically significant and even fewer adducts are highly specific and relevant to cancer induction.

The overall correlation of short-term tests with carcinogenic potential of various compounds is estimated to be approximately 60 per cent, and application of a battery of short-term tests did not improve the correlation. However, the chemicals that cause cancer by reacting with DNA can be identified with some precision in short-term mutagenicity studies. Because non-genotoxic carcinogens exist, and because level of knowledge about the mechanisms of carcinogenesis is relatively incomplete, short-term genotoxicity testing for carcinogens cannot be a substitute for long-term testing.

Since many human carcinogens initiate cancer by means other than reaction with DNA (e.g., non-genotoxic agents which may induce cancer by processes such as stimulation of growth of dormant initiated cells, an action generically referred to as `promotion'), short-term mutagenicity tests (e.g., point mutations, chromosome aberrations, germ cell mutations, SCE, and DNA adduct formation) are incapable of identifying them. To uncover such carcinogens, short-term tests for tumour promotors (e.g., cell-to-cell communication, cell conductivity) would be highly relevant.

Cancer is the second leading caue of death in industrialized countries. Epidemiological evidence (as demonstrated by different cancer patterns in various parts of the world and by the changing cancer patterns in migrants) exists indicating that environmental factors (including lifestyle) play a major role in cancer.

Pesticides are used widely. Some small populations receive rather high doses during the manufacture and application of pesticides; by contrast, large populations receive relatively low doses via residues in food, and by contamination of tap water from farm runoff and leaching from hazardous waste sites. Several pesticides produce tumours in animals, and so concern is widespread that the use of some pesticides may cause cancer in humans.

Epidemiological studies provide little evidence about the carcinogenicity of pesticides in humans, because suitable study populations are difficult to find. Except for arsenic, the dearth of direct evidence demonstrating that pesticides do or do not cause cancer in humans reflects in part a lack of adequate studies and the difficulties of epidemiological studies in establishing causal relationships. Therefore, claims that pesticides are or are not carcinogenic in humans are to be viewed with caution.

Cancer is a complex process, and its causes are poorly understood (Pitot, 1986). Years of observations have confirmed that cancer generally arises from single cells whose genetic structure and function have been irreversibly altered, with the resultant loss of control of normal growth. The nature of this genetic damage is poorly understood, although considerable progress has been made in recent years to clarify the role of oncogenes and antioncogenes in neoplastic transformation.

Cancers differ so markedly in their characteristics of differentiation, growth rate, and local and distant (metastatic) invasiveness, that one can only conclude that the normal growth control mechanisms must be exceedingly complicated. Knowledge of these changes is in a relatively primitive state. Cancer is also a progressive disease, such that tumours usually tend to evolve with time toward higher grades of malignancy at times after cessation of exposure.

This concept has led to the classification of carcinogenesis process into initiation, promotion, and progression. Some substances are complete carcinogens, having the properties of all three.

Carcinogens differ in the nature of their molecular interactions. So-called genotoxic carcinogens are either electrophilic per se (i.e., direct-acting) or are metabolized to an electrophilic form which permits them to interact with nucleophiles in cells, especially DNA. The sites of interaction of electrophiles with DNA are determined by complex chemical factors, which, in turn, determine the type and likelihood of mutations. DNA repair processes are important modifiers of the amount and duration of damage.

The complexity of carcinogenesis is indicated by the number of dissimilar classes of carcinogens: for example, polycyclic aromatic hydrocarbons, aromatic amines, nitroso compounds, direct-acting alkylating and cross-linking agents, metals, fibres, and hormones.

Although electrophilic or genotoxic carcinogens are the dominant class of known human carcinogens, a substantial number of agents are non-genotoxic, and yet have been shown to produce tumour in animals. Promoting agents generally fall into this class. Although there has been a general reluctance to regard genotoxic carcinogens as having a threshold response, there is a growing tendency to regard at least some non-genotoxic carcinogens as having a reversible mode of action and as more likely to have a threshold for cancer induction. The linear non-threshold doseresponse relationship is generally adopted for genotoxic carcinogens.

The following approaches are used to determine whether a chemical causes cancer or can be judged as likely to cause cancer in a specified species. The approaches include structureactivity analysis, mutagenicity tests, lifetime exposures of laboratory animals, and human epidemiological investigations. Taken together, the findings of these analyses provide the basis for judging the scientific strength of conclusions regarding carcinogenicity and for applying the findings to the quantitative estimation of the magnitude of cancer response.

This approach attempts to link chemical structure with carcinogenic activity (Ennever and Rosenkrantz, 1989). In some cases, the relationship can be simple, as with direct acting alkylating agents. For instance, the powerful carcinogen bis-chloromethyl ether (BCME) was identified as a probable carcinogen before it had been studied in animals or exposed chemical workers on the basis of its being a bi-functional alkylating agent with cross-linking properties. In general, the correlation of chemical structure with mutagenic activity is much more highly developed than with carcinogens; however, a strong correlation exists between carcinogenesis and mutagenesis with genotoxic carcinogens. The structure-activity approach is inexpensive and rapid; however, it is insufficiently developed at present to be reliably used on a routine basis.

A detailed discussion of these relatively inexpensive, rapid, and widely used assays is found above. Genotoxic agents are very likely to be carcinogenic; however, the results give no reliable estimate of carcinogenic potency. Negative results of genotoxicity assays do not rule out conclusively the possibility that a substance may cause cancer (Tenant et al., 1987).

Tumour induction in laboratory animals is the most relevant biological response to evaluate the carcinogenicity for humans (Sontag et al., 1976). Previously, a standardized study was commonly performed under the following conditions: two doses (the current preference is three or four); both sexes; two species of rodent (e.g., F-344 rats and B₆C₃F₁ mice); dosing for 2 years in rats and 18 months in mice; and administration by gastric intubation, incorporation into feed or drinking water, or inhalation. The route of exposure was selected to closely simulate human exposure. These tests are quite costly (each about $500 000 (US)), and require almost five year to complete; consequently, only a limited number of agents can be tested.

Some responses in animals are difficult to relate to humans. Such questionable responses, for example, include the increased incidence of tumours at sites absent in humans, a shortening of the time before tumours occur with relatively high frequency, and the induction of mostly benign tumours unlikely to progress to malignancies. Nevertheless, for all their limitations, long-term animal studies are the mainstay of chemical carcinogen testing.

This discipline is the most accurate indicator of human carcinogenicity, particularly when positive results are obtained consistently in multiple studies under various exposure conditions and with suitable control over confounding factors (Kalder and Day, 1985). However, epidemiology is relatively insensitive; even under ideal circumstances, it is incapable of detecting less than a 20 per cent increase in cancer. Stringent requirements exist for the execution of rigorous population studies: primarily, an adequate number of exposed subjects with appopriate controls are required, and both groups must be followed long enough to permit the expression and detection of carcinogenicity. Both positive and negative epidemiological studies are useful in quantitative risk assessment, since the negative studies set upper limits on a response. Because the number of agents studied adequately in humans is very limited, the lack of human evidence should not generally be taken as an indication of an absence of carcinogenicity.

A number of other tests have also been used occasionally to detect carcinogenicity; some are promising though still in the developmental stage. Initiation-promotion studies on mouse skin and in rat liver are used on a selective basis to determine whether an agent has promoting action. Since cancer promotion is organ-specific, other initiationpromotion models used in a limited way include mouse forestomach, bladder, intestine, and mammary gland. Even if the results of these tests were unequivocal, the interpretation of the findings remains complex and controversial. For example, dioxin* produces an excess tumour response in a standard long-term animal feeding study; it also shows strong promoting activity in the rat liver promotion assay. Does this mean that dioxin is only a promotor or could it be a whole carcinogen with strong promoting activity

Other tests for promoting activity being developed include mammalian cell transformation assays and assays of cell-to-cell communication.

If a substance such as a pesticide is found to cause cancer, the estimation of cancer risks in humans can be undertaken using a framework promulgated by the US National Academy of Sciences (1983) and incorporating four steps that facilitate the analysis: hazard identification, doseresponse assessment, exposure assessment, and risk characterization.

Hazard identification deals with the question of how likely the agent is to be a human carcinogen. A weight-of-evidence approach is used to evaluate the quality and quantity of evidence for carcinogenicity. Evidence from studies in rodents and humans are weighted most; but results of structureactivity relationships, short-term genotoxicity assays, and metabolic interactions with DNA are also relied upon. Similar weight-of-evidence schemes are employed by IARC and the US EPA (1986). There is a growing tendency to use alternative regulatory strategies for various grades of evidence.

Estimation of risks below the observed levels of response in animals or humans requires some mathematical model to extrapolate doseresponse curves below the observation range. The dominant one in carcinogen risk assessment is the conservative linear non-threshold doseresponse model, which implies that there is no risk-free dose above zero. However, the shape of the dose-response relationship is not verifiable by direct observation, and its plausibility depends on an understanding of the relevant aspects of the underlying carcinogenic mechanisms. The use of low-dose extrapolation is particularly uncertain for promotors that show evidence of reversible action. Mathematical models which depend on time patterns of tumour occurrence exist, and are being used to limited extend in dose-response assessment.

Exposure assessment is one of the major sources of uncertainty in risk assessment with respect to ambient concentrations of pesticides to which individuals are exposed. Toxicokinetics are employed to define target tissue dose, although they are often hampered by a lack of understanding of which metabolite is the aetiologic agent.

Risk characterization combines the integrated dose† with the cancer potency value to obtain estimates of cancer risk as either excess numbers of cancers in an exposed population or excess lifetime risk to an exposed individual. The weight-of-evidence conclusion about carcinogenicity should be included with the risk estimates.

*Dioxin is the abbreviated name for 2,3,7,8-tetrachlorodibenzo-p-dioxin or TCDD.
†For example, the lifetime-average-daily-does or LADD.

The risk estimates can be used for risk management solutions that may entail the value-laden determinations of acceptability of risks or the weighting of incommensurables such as risks versus benefits or lives versus financial profits.

As a result of chronic exposure to chemicals, the liver may be injured in a variety of ways, such as degeneration, cell proliferation, necrosis, and induction or inhibition of various enzyme activities. The accumulation of lipids in a degenerated liver may have largely undesirable toxic effects in the organ. The herbicide metribuzin is known to cause hepatotoxicity through the process of depletion of glutathione from the liver.

Continued prolonged exposure to pesticides or herbicides at sufficiently high doses can result in accumulation of fat in the liver. This accumulation of lipids may also provide a storage site for these compounds. The accumulation of pesticides and lipid in the liver may result in the alteration of liver function, cell proliferation, necrosis, or some secondary effects which may be manifest as a variety of pathological consequences. In many cases, the damage in the liver results from biotransformation of the parent compounds to toxic metabolites, some of which can be potent mutagens or carcinogens.

Pesticides often undergo metabolic transformation in the liver; for example, DDT is converted to DDE or to DAA by different pathways, and the combined toxicity of the metabolites is considerably less than that of the parent compound (DDT). The observed hepatic tumour formation by chlorinated pesticides is generally caused by effects of metabolites rather than of the parent compound.

Two classes of methods are available to detect chronic toxicity from these substances in the liver: structural and functional. The structural methods involve histological examination of tissues by microscopic techniques specifically developed to measure the structural or compositional changes associated with exposure to these substances. The sensitivity of these procedures is usually limited by the availability of microscopy equipment and trained histopathologists.

In the liver, alkyl pesticides are known to induce a large number of enzymes, such as alkaline phosphatase, that can serve as indicators of biological responses when observing the toxicity of these substances. Key enzyme systems include cytochrome P₄₅₀ monooxygenases, phosphoesterases, glutathione-S-transferases, and O-alkyl and O-aryl conjugation.

3.3 REFERENCES

Amann, R. P. (1982) Use of animal models for detecting specific alterations in reproduction. Fund and Appl. Toxicol. 2, 13-26.

Baginsky, E. (1982) Occupational Disease in California, Division of Labour: Statistics and Research, California Department of Industrial Relations, San Francisco.

Chernoff, N. and Kavlock, R. J. (1982) An in vivo teratology screen utilizing pregnant mice. J. Toxicol. Environ. Hlth. 10, 541-550.

Durham, W. F., Wolfe, H. R. and Elliot, J. W. (1972) Absorption and excretion of parathion by spray men. Arch. Environ. Health 24, 381-7.

Echobichon, D. J., Hansell, J. and Safe, S. (1977) Halogen substituents at 4 and 4' positions of biphenyl influence on hepatic function in rat. Toxicol. Appl. Pharmacol. 42, 359-66.

Ennever, F. K. and Rosenkrantz, H. S. (1989) Application of the carcinogenicity prediction and battery selection method to recent national toxicology program shortterm test data. Environ. Molec. Mutagen. 13, 332-338.

Feldman, R. J. and Maibach, H. I. (1974) Percutaneous penetration of some pesticides and herbicides in man. Toxicol. Appl. Pharmacol. 28, 126-132.

Franklin, C. A. (1984) Estimation of dermal exposure to pesticides and its use in risk assessment. Can. J. Physiol. Pharmacol. 62, 1037-39.

Hardin, B. D., Becker, R. A., Kavlock, R. J., Seidenberg, J. M. and Chernoff, N. (1987) Overview and summary: workshop on the Chernoff/Kavlock preliminary developmental toxicity test. Teratog. Carcinog. Mutagen. 7, 119-127.

Kalder, J. and Day, N. (1985) The use of epidemiological data for the assessment of human cancer risk. In: Hoel, D. G., Merrill, R. and Perera, F. P. (eds), Risk Quantitation and Regulatory Policy, Banbury Report 19, Cold Spring Harbor Laboratory.

Knox, J. M., Tucker, S. B. and Flannigan, S. A. (1984) Paraesthesia from cutaneous exposure to a synthetic pyrethroid insecticide. Arch. Dermatol. 120, 744-746.

Morison, W. L., Parrish, J. A. and Woehler, M. E. (1981) The influence of ultraviolet radiation on allergic contact dermatitis in the guinea pig: 1 UVB radiation. Br. J. Dermatol. 104, 161-4.

Morrissey, R. E., Schwetz, B. A., Lamb, J. C. IV, Ross, M. D., Teague, J. L. and Morris, R. W. (1988) Evaluation of rodent sperm, vaginal cytology, and reproductive organ weight data from National Toxicology Program 13-week studies. Fund. and Appl. Toxicol. 11, 343-358.

National Academy of Sciences (1983) Risk Assessment in the Federal Government, National Academy Press, Washington, DC.

Schardein, J. L. (1985) Chemically Induced Birth Defects, Marcel Dekker, New York.

Schardein, J. L., Schwetz, B. A. and Kenel, M. F. (1986) Species sensitivities and prediction of teratogenic potential. Environ. Hlth. Perspect. 61, 55-67.

Sontag, J. M., Page, M. P. and Saffiotti, U. (1976) Guidelines for carcinogen bioassays in small rodents. Carcinogenesis Program, National Cancer Institute, DHEW Publication 16 (NIH) 76-801, 65pp.

Tennant, R. W., Margolin, B. H., Shelby, M. D., Zeiger, E., Haseman, J. K., Spalding, J., Caspary, W., Resnick, M., Stasiewicz, S., Anderson, B. and Minor, R. (1987) Prediction of chemical carcinogenicity in rodents from in vitro genetic toxicity assays. Science 236, 933-941.

US Environmental Protection Agency (1986) Guidelines for carcinogen risk assessment. CFR 51, 33992-34003.

Working, P. K. (1988) Male reproductive toxicology: comparison of the human to animal models. Environ. Hlth Perspect. 77, 37-44.

Yonemoto, K., Gellin, G. A. and Epstein, W. L. (1983) Reduction in eumelanin by activation of glutathione reductase and gamma glutamyl transpeptidase after exposure to a depigmenting chemical. Biochem. Pharmacol. 32, 1379-82.

3.1 ACUTE TOXICITY
	3.1.1 MORBIDITY AND MORTALITY: MAGNITUDE OF THE PROBLEM
		3.1.1.1 Factors contributing to morbidity and mortality
		3.1.1.2 Methods for investigation and surveillance of morbidity and mortality
	3.1.2 EPIDEMIOLOGICAL STUDIES
	3.1.3 TESTS FOR ACUTE TOXICITY
		3.1.3.1 Tests for pesticide neurotoxicity
		3.1.3.2 Tests for immunotoxicity
	3.1.4 SKIN ABSORPTION OF PESTICIDES AND ACUTE POISONING
		3.1.4.1 Clinical forms of acute exposure of skin to pesticides
		3.1.4.2 Methods to assess acute skin exposure
	3.1.5 OTHER TARGET ORGANS OF ACUTE TOXICITY
	3.1.6 CONCLUSIONS AND RECOMMENDATIONS
3.2 CHRONIC TOXICITY
	3.2.1 REPRODUCTIVE AND DEVELOPMENTAL TOXICITY
		3.2.1.1 Animal tests for reproductive toxicity
		3.2.1.2 Animal tests for developmental toxicity
		3.2.1.3 Extrapolation from animals to humans
	3.2.2 GENOTOXICITY
		3.2.2.1 In vitro assays for genetic damage
		3.2.2.2 In vivo assays of genetic damage
		3.2.2.3 Carcinogenic risk assessment using results of short-term tests
	3.2.3 CHEMICAL CARCINOGENESIS
		3.2.3.1 Tests to identify chemical carcinogens and prospective carcinogens
		3.2.3.2 Risk assessment methodology
	3.2.4 TARGET ORGAN TOXICITY: LIVER
	3.2.5 CONCLUSIONS AND RECOMMENDATIONS
3.3 REFERENCES