Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Developmental Effects of Chemical Contaminants The discussion in this chapter is limited to embryo (fetal) death, growth retardation, and malformations the only end points measured in the Food and Drug Administration's (FDA) guidelines for Segment II develop- mental toxicity studies of drugs (Collins, 1978) and, therefore, the only end points for which there are sufficient data bases for analysis. Postnatal functional impairment is not covered, despite its relevance, since there is no well-established data base with which to make cross-species compar- isons. The much broader spectrum of end points included under the head- ing of reproductive toxicology, e.g., germ cell toxicity, infertility, and dysfunction of the adult reproductive system, are covered in Chapter 3. Embryo lethality is defined and reported in the literature as the ratio of resorptions or dead fetuses in a litter at term to the number of implantation sites. Growth retardation is measured by weighing and taking crown-to- rump measurements of live fetuses at term. The frequency and type of structural anomalies are determined by gross inspection of the fetuses and by detailed skeletal and soft tissue analysis. The occurrence of embryo death precludes measurements of growth retardation or identification of malformations, because these two end points are noted only on live fetuses. Developmental toxicity includes any detrimental effect produced by exposures during embryonic stages of development. Such lesions can be either irreversible or reversible. Embryolethal lesions result in resorption, spontaneous abortion, or stillbirth. Persistent lesions that cause overall growth retardation or delayed growth of specific organ systems are gen- erally referred to as embryotoxic. For a chemical to be labeled a teratogen, it must significantly increase the occurrence of irreversible structural or 11
|2 DRINKING WATER AND HEALTH functional abnormalities in live offspring after it is administered to either parent before conception, to the female during pregnancy, or directly to the developing organism. Many teratologists believe that any chemical administered in appropriate dosages at certain developmental stages can cause some disturbances in embryonic development in some laboratory species (Fabro et al., 1982; Karnofsky, 1965; Staples, 19751. For an agent to be classified as a de- velopmental toxicant, it must produce adverse effects on the conceptus at exposure levels that do not induce severe toxicity in the mother (e.g., substantial reduction in maternal weight gain, persistent emesis, hypo- or hyperactivity, or convulsions), so that the effects are not secondary to the stress on the maternal system. The main reason for conducting develop- mental toxicity studies is to ascertain whether an agent causes specific or unique toxic effects on the conceptus. If these studies are conducted under extreme conditions of maternal toxicity, then identification of exposures uniquely toxic to the conceptus or pregnant animal is not possible. (This is discussed in more detail later in this chapter.) In some cases, however, chemical agents are deliberately administered at maternally toxic doses to determine the threshold level for adverse effects on the offspring. As a result, conclusions can be qualified to indicate that adverse effects on the conceptus were obtained at maternally toxic exposure levels and may not be indicative of selective or unique developmental toxicity. INFLUENCE OF TIME OF EXPOSURE Compared to adults, developing organisms undergo rapid and complex changes within a relatively short period. Consequently, the susceptibility of the conceptus to chemical insult varies dramatically within each of the major developmental stages, i.e., the preimplantation, embryonic, fetal, and neonatal stages. As shown in Table 2-1, the time between ovulation and preimplantation development is similar among several mammalian species, regardless of gestation length (Brinster, 1975~. Alterations in the hormonal milieu as well as direct secretion of chemicals into uterine fluids during this period can interfere with implantation and result in embryo death. The preimplantation embryo appears to be more susceptible to death than to teratogenicity following chemical insult. In studies with preim- plantation embryo cultures, severe toxicity was manifested by rapid death of the embryo, and the less severe effects included decreased cleavage rates and arrested development (Brinster, 1975~. There have been few studies on the effects of sublethal exposures to preimplantation embryos, and the possibilities of persistent biochemical or morphological alterations have not been adequately explored.
Developmental Effects of Chemical Contaminants 13 TABLE 2-1 Timing of Early Development in Some Mammalian Speciesa Times of Early Development (days from ovulation) Length of Blastocyst Gestation Mammal Formation Implantation Organogenesis (days) _ . . . Mouse 3-4 4-5 6-15 21 Rat 3-4 5-6 6-15 22 Rabbit 3-4 7-8 6-18 30 Sheep 6-7 17-18 14-36 150 Monkey (rhesus) 5-7 9-11 20-45 164 Human 5-8 8-13 21-56 270 aAdapted from Br~nster, 1975. Following implantation, organogenesis takes place. During that period, there are highly specific periods of vulnerability for different organ sys- tems, thus making the embryo extremely susceptible to the induction of structural birth defects. The periods when the major embryonic organ systems of the rat are most sensitive to teratogenic insult are shown in Figure 2-1. Administration of a teratogen on day 10 of rat gestation is likely to result in a high level of brain and eye defects, intermediate levels of heart and skeletal defects, and a low level of urogenital defects. If the same agent were administered on day 11, a different spectrum of mal- formations would be anticipated, predominantly effects on the brain and palate. Figure 2-1 also illustrates that exposure to teratogens usually results in a spectrum of malformations involving a number of organ systems, reflecting the overlap of critical periods for individual organ systems. This is most evident in species such as rodents, which have short gestation periods, but can also be observed in humans. Most teratogens have been found to influence the development of several organ systems in humans and to cause clusters of malformations rather than single anomalies. The critical period of inducing anomalies in individual organ systems may be as short as 1 day or may extend throughout organogenesis. In the rat, for example, urogenital effects can be produced by drug treatment from the 9th to 18th day of gestation. This implies that development of the urogenital system is multiphasic and that individual stages may have different sensitivities to chemical insult. Depending on the mechanism of action of the agent and the time of administration, it is possible that only one or a few of these steps will be affected but that succeeding stages will be disrupted as a result of the original alteration. The persistence of the agent also influences the malformation pattern, as discussed later in
4 DRINKING WATER AND HEALTH 50 z ° 40 z a: ' 30 o o 20 z - ~ 10 o J a: o ~1 / - I'' i// 1,- ~ Brain I ~ \ /Heart and \ \ / Axial Skeleton \ Palate Aortic \, ~_'\ Arches _ _ ~ ~ ~ < ,_ - _ I ~~-t- 9 10 11 Urogenita I ""_ ~ "~_ _` 1 - 1 8 12 13 14 DAYS OF GESTATION IN THE RAT 15 16 FIGURE 2-1 Hypothetical pattern of the susceptibility of rudimentary embryonic organs to teratogenic insult. Adapted from Wilson, 1965. this chapter. Processes governing embryonic differentiation are not well understood, yet they are most likely to determine the intrinsic susceptibility of individual organ systems to teratogenic insult. Histogenesis, functional maturation, and growth are the major processes occurring during the fetal and neonatal (i.e., perinatal) periods. Insult at these later developmental stages leads to a broad spectrum of effects that can generally be manifested as growth retardation, functional disorders, or transplacental carcinogenesis. The fetus is more resistant to lethal effects than is the embryo, but the incidence of stillbirths is measurable. The perinatal period is a time of high susceptibility to carcinogenesis. At least three factors contribute to this enhanced susceptibility: high cellular rep- lication rates, ontogeny of xenobiotic-metabolizing enzymes, and low immunocompetence. Several childhood tumors occur so early in life that prenatal origin is considered likely. Among these are acute lymphocytic (but not myelogenous) leukemia, Wilms' tumor, neuroblastoma, carci- noma of the liver, and presacral teratoma (Miller, 1973~. In 1976 cancer was the chief cause of death from disease among children under the age of 15 in the United States, accounting for 11.3% of all childhood deaths. Leukemias and lymphoma accounted for approximately half of these deaths, followed by cancers of the central nervous system, soft tissues, kidney, and bone (ACS, 19801. Studies of direct-acting transplacental carcinogens, such as ethylnitro- sourea (ENU), indicate that susceptibility to carcinogens in rodents begins
Developmental Effects of Chemical Contaminants ~ 5 after completion of organogenesis. In one study, tumors in offspring oc- curred primarily when ENU was given during the fetal period, whereas birth defects and embryo deaths predominated when exposures were ad- ministered earlier in organogenesis (Ivankovic, 19791. This is not to imply that teratogenesis and carcinogenesis are mutually exclusive processes, however. Birth defects and neoplasms occur together in the same offspring with unusually high frequency, but not necessarily at the same site. Ter- atogenesis and carcinogenesis can be considered graded responses of the embryo to injury, teratogenesis representing the grosser response involving major necrosis. Bolande (1977) has postulated that certain agents cause teratogenesis in early, relatively undifferentiated embryos; combined car- cinogenesis and teratogenesis when older embryos are exposed; and car- cinogenesis alone when exposure occurs during the perinatal period. Prenatal insult may also predispose the offspring to tumor induction in later life. PATTERNS OF DOSE RESPONSE IN LABORATORY ANIMAL STU Dl ES Functional deficits and perinatally induced cancers, such as those caused by diethylstilbestrol (DES) (Herbst et al., 1977), are often not manifested until adolescence or later. They are usually examined as end points in themselves without correlation to outcomes observable at the time of birth. Observations made at the time of birth indicate that the major effects from prenatal exposure are embryo death, malformations, and growth retar- dation. The relationship between these outcomes is complex, and varies with the agent, the time of exposure, and the dose. Some developmental toxicants can cause malformations in the entire litter at exposure levels not causing embryo death. The dose-response pattern for such agents is shown in Figure 2-2A. If the dose is increased beyond that causing malformations of the entire litter, embryo death can occur, but often in conjunction with maternal toxicity. Fetal malformations are usually accompanied by growth retardation. Note that the curves for these two end points are parallel and slightly displaced from one another in Figure 2-2A. This pattern of response is rare, indicating that the agents have high teratogenic potency. Both natural and synthetic glucocorticoids cause this kind of dose-response pattern. The target-organ specificity of glucocorticoids is related to the concentration of glucocorticoid receptor protein, which is higher in the craniofacial region than in other parts of the embryo (Pratt and Salomon, 19811. Thus, pharmacological doses ad- ministered to laboratory animals at midgestation induce malformations of the palate. Glucocorticoid induction of cleft palate in the absence of other major malformations, embryo death, extensive necrosis, or growth retar-
6 DR! N KING WATER AND H "LTH loo Malformation / / Growth / Retardation At// o ~ / Lethal it`,r A On LL t 1oo LL 7 o CL US o Ct: o 100 o /:th,al itv / Growth / / Retardation >/ ~ '.' Ma If ormation B / Letha I ity Growth / Retardation ~ / 1 ~ / C EMBRYOTOX IC RANGE OF DOSES ~ FIGURE 2-2 Theoretical dose-response patterns for different types of embr~rotoxic agents. Adapted from Neubert et al., 1980. cations is an example of developmental toxicity with selective teratogenic potency. A more common dose-response pattern involves a combination of re- sorptions, malformations, growth retardations, and unaffected fetuses after exposure to a developmentally toxic range of doses of an agent (Figure 2-2B). Lower doses may cause predominantly resorptions or malforma- tions, depending on the teratogenic potency of the agent. As the dosage increases, however, embryo death predominates until the entire litter is resorbed. Growth retardation can precede both outcomes or parallel the malformation curve. This response pattern is typical of agents that are cytotoxic to replicating cells by alterating replication, transcription, trans- lation, or cell division. These agents include alkylating, antineoplastic, and many mutagenic substances. The susceptibility of the embryo to these agents derives from the high rate of cell division during organogenesis.
Developmental Effects of Chemical Contaminants 17 Low doses of cytotoxic agents administered relatively early in the critical period may kill cells at rates low enough that the cells can be replaced through compensatory hyperplasia, resulting in growth-retarded but mor- phologically normal fetuses at term. Higher doses administered later during the critical period may substantially deplete cell number, leaving insuf- ficient time for replacement before critical morphogenetic events occur. The resulting hypoplasia of the rudimentary organs and retarded prolif- eration of surviving cells are the initial events leading to the induction of malformations. High levels of exposure may damage too many cells and organ systems to be compatible with survival, thus resulting in embryo death (Ritter, 19771. Exposure to cytotoxic agents during organogenesis can produce all three outcomes both within and among litters. Some litters may be totally resorbed, others may include only growth-retarded fetuses at term, and still others may include a mixture of malformed or growth- retarded fetuses and resorption sites. A third dose-response pattern consists of growth retardation and embryo death without malformations (Figure 2-2C). The dose-response curve for embryo death in this case is usually steep, which may imply a dose threshold for survival of the embryo. Growth retardation of surviving fetuses usually precedes a significant increase in lethality. Agents pro- ducing this pattern of response would be considered embryotoxic or em- bryolethal substances but not teratogenic. When such a pattern is observed, it is necessary to conduct additional studies with doses within the range causing growth retardation and embryo death. Results obtained at these intermediate doses can indicate whether teratogenicity has been masked by the deaths of the embryos. Agents in this class include the mitochondrial protein synthesis inhibitors chloramphenicol and thiamphenicol (Neubert et al., 1980~. On days 10 and 11 of treatment with thiamphenicol, the dose-response curve for embryo death is steep, increasing from baseline to 100% mortality at doses between 100 and 125 mg/kg body weight per day (Bass et al., 19781. In the same study, dose-dependent inhibition of mitochondrial respiration, adenosine triphosphate (ATP) content, and cy- tochrome oxidase activity in embryonic tissue was correlated with growth retardation and death of the embryos. There is no basis for target-organ susceptibility to perturbation of such fundamental cellular processes in the early embryo. Consequently, all tissues appear to be equally affected. An early sign of perturbation is overall growth retardation, which progresses to death of the entire litter once a threshold for cellular energy requirements is exceeded. These conditions are incompatible with teratogenicity, which can induce irreversible lesions in some tissues while sparing others, thus permitting survival of abnormal embryos to term. For some agents, i.e., those cytotoxic to replicating cells (Figure 2- 2B), growth retardation, embryo death, and teratogenicity are viewed as
i~ DRINKING WATER AND HEALTH different degrees of manifestation of the same primary insult, cytotoxicity. For others, there is a qualitative difference in response, and the primary insult leads to embryo death alone (Figure 2-2C) or to teratogenicity alone (Figure 2-2A). Separate evaluations of growth retardation, teratogenicity, and embryo death for increasingly higher doses are necessary to determine the agent's primary mode of action. In safety studies, the usual sequence of testing begins with dose range- finding studies in relatively small numbers of pregnant rodents. On days 6 through lS of gestation, animals are exposed to the test agent at doses up to and including those causing limited maternal toxicity or develop- mental toxicity (e.g., death or severe growth retardation). The purpose of this type of study is to obtain a qualitative yes-or-no signal about the potential developmental toxicity of the agent. At the next level of testing, larger numbers of animals are exposed on days 6 to 15 of gestation to obtain quantitative information on dose-response relationships. The high- est dose should cause measurable maternal toxicity (e.g., significant depression of weight gain) or developmental toxicity (e.g., significant depression of fetal body weight or increased embryo death), and the low dose should cause no observable effects. If evidence of selective devel- opmental toxicity is obtained from this study, it may be necessary to conduct a third study, exposing dams on single days during organogenesis at doses that are not maternally toxic, to obtain a clear definition of the dose-response pattern of developmental toxicity. EXTRAPOLATION OF ANIMAL DATA TO HUMANS Spectrum of End Points The timing of exposure and the patterns of dose response obtained in animal studies have important implications for extrapolating the resultant data to humans. The major implication is that a spectrum of end points can be produced, under the controlled conditions of timing and exposure that can be achieved in animal studies. In some cases, the spectrum comprises a continuum of response: depressed birth weight or functional impairment at low doses, birth defects at intermediate doses, and death at high doses. Less commonly, birth defects alone or deaths alone are produced. Consequently, in estimating risks to humans, all exposure- specific adverse outcomes must be taken into consideration not just birth defects. Most often neglected in extrapolation of animal data to humans is fetal growth retardation, despite the strong evidence concerning the adverse consequences of low birth weight in human infants (Hull et al., 19781. Fetal growth retardation in the absence of a significant reduction
Developmental Effects of Chemical Contaminants 19 TABLE 2-2 Frequency of Selected Adverse Pregnancy Outcomes in Humansa Event Frequency per 100 Pregnancies 10-20 Spontaneous abortions, 8 to 28 weeks Chromosome anomalies in spontaneous abortions, 8 to 28 weeks Chromosome anomalies detected by amniocentesis Stillbirths Low birth weight (<2,500 g) among live births Major malformations among live births Chromosome anomalies among live births Severe mental retardation among children <15 years old 30-40 2-4 2-3 0.2 0.4 aAdapted from Edmonds et al., 1981. in maternal weight gain is an important event to be considered in cross- species extrapolation. A similar spectrum of response has been observed in humans after prenatal exposure to developmental toxicants. The specific effects in that spectrum are determined by the time and duration of exposure, magnitude of exposure, interindividual differences in sensitivity, interactions with other types of exposure, and interactions among all these factors (Fraser, 1977~. Consequently, manifestations of developmental toxicity cannot be presumed to be constant or specific across species; i.e., an animal model cannot be expected to forecast exactly the human response to a given exposure. For instance, an agent that induces cleft palate in the mouse may elevate the frequency of spontaneous abortion or intrauterine growth retardation in humans. Any manifestation of exposure-related develop- mental toxicity in animal studies can be indicative of a variety of responses in humans (Kimmel et al., 19841. Table 2-2 illustrates another factor to be considered in cross-species extrapolation. The most common adverse pregnancy outcome in humans is spontaneous abortion or early fetal loss (before the 28th week of preg- nancy), occurring in at least 10% to 20% of all recognized pregnancies. Estimates from prospective studies range even higher: between 20% and 25% of all conceptions spontaneously abort (Edmonds et al., 19811. The incidence of spontaneous abortions is high during early pregnancy, es- pecially during the first 12 weeks, and gradually decreases to the 20th
20 DRINKING WATER AND H"LTH week, after which fetal loss is uncommon. Approximately one-third of the specimens obtained from spontaneous abortions occurring between 8 and 28 weeks of gestation contain chromosome aberrations. The frequency of such aberrations is at least 60-fold higher among spontaneous abortions than among term births. Among the spontaneous abortions without chro- mosome aberrations, approximately half have structural malformations (Edmonds et al., 1981~. The frequency of such malformations is not as well documented as that of chromosome aberrations, because they are difficult to observe in specimens that are often macerated or incomplete. In the remaining one-third of the specimens, the incidence of placental inflammations suggests that uterine infections can be high (Ornoy et al., 19811. These observations suggest that the majority of human embryos bearing chromosome aberrations or morphological abnormalities are lost through early miscarriage. Epidemiological approaches to monitoring the fre- quency of early fetal loss and detecting such fetal abnormalities have only been used to a limited extent. Most studies of humans focus on the ex- amination of adverse effects, such as major malformations, stillbirths, low birth weight, and neonatal deaths, at the time of birth or later. Underes- timation of adverse pregnancy outcome, and thus true risk and pattern of response, is unavoidable in human studies whenever measurements are made only from the time of birth onward. Moreover, it is difficult to design studies to work within the limitations of statistical power and to document exposure of humans, even when observations are confined to the time of birth onward (Edmonds et al., 1981; IRLG Epidemiology Work Group, 19811. Developmental toxicants with dose-response patterns resembling those in Figure 2-2A could be detected by monitoring malformations at the time of birth, especially if the malformations were rare (such as those resulting from thalidomide), or if the exposed populations were large (such as those with rubella infections). The possibility of concordance in the pattern of malformation across species would be greatest for potent teratogens op- erating in the pattern shown in part A of Figure 2-2, because they tend to demonstrate target-organ specificity. Agents with patterns shown in parts B and C would\probably be missed, because early fetal loss is not routinely monitored in human populations, even though it has been done successfully in isolated groups (Kline et al., 19771. Concordance of Results from Animal and Human Studies For several well-studied developmental toxicants, there is good evidence for dose-response correspondence between humans and animals. The cor- respondence is nearly 100% when data on animals have been expected to
Developmental Effects of Chemical Contaminants 2 ~ provide a qualitative yes-or-no signal, i.e., when any exposure-related adverse effect from an animal study is taken into consideration and not just those specific outcomes that are also seen in humans. Thalidomide is the only toxicant known to produce developmental abnormalities in humans but not to produce such effects consistently in conventional lab- oratory animal species (see reviews by Brent, 1972; FDA, 1980; Frankos, 1985; Fraser, 1977; Nisbet and Karch, 1983; Nishimura and Tanimura, 1976; Schardein, 1976; Shepard, 1980; Strobino et al., 1978; and Wilson, 1973~. An FDA review of the literature (FDA, 1980; Frankos, 1985) indicates that of 38 compounds having demonstrated or suspected tera- togenic activity in humans, all except one tested positive in at least one animal species. Furthermore, more than 80% were positive in more than one species. Eighty-five percent of the 38 compounds were teratogens in mice, 80% in rats, 60% in rabbits, 45% in hamsters, and a low of 30% in primates. Other species (i.e., the cat, ferret, and guinea pig) have been used to test only a few of these substances. Nisbet and Karch (1983) reported that humans appear to be as sensitive to thalidomide as the most sensitive species tested (the cat), and 5 to 10 times more sensitive than those species with comparable target-organ specificity for limb defects (the rabbit and various primates). These authors also compared the minimally effective doses in animals and humans of eight known teratogens where there was cross-species concordance in target organs. When dose was converted to mg/kg body weight per day for these teratogens fi.e., thalidomide, polychlorinated biphenyls (PCBs), alcohol, aminopterin, methotrexate, methylmercury, DES, and diphenyl- hydantoin], humans were shown to be more sensitive than laboratory animals by factors ranging from 1.8 (for PCBs) to 50 (for methylmercury). If exposure was expressed in units of dose per unit body surface area per day, the ratios of human to animal sensitivity ranged from 0.3 to 8.0 (Nisbet and Karch, 19831. These findings of qualitative (yes-or-no signals) and quantitative (dose- response) concordance support the use of animal studies for predicting risk to humans. However, there are qualifications that must be placed on their direct application to risk estimation. These comparisons of response have necessarily been limited to agents for which there have been estab- lished effects in humans and substantial data bases from animal studies. Only the eight agents tested could be found to meet these criteria (Nisbet and Karch, 19831. In the Catalog of Teratogenic Agents, Shepard (1980) has listed more than 600 agents that cause congenital anomalies at any dose in laboratory animals. Only 20 of these are confirmed or suspected developmental toxicants in humans. Consequently, the major concern in risk assessment today is that far more agents have been shown to be positive in animal studies than have been identified in human studies for
22 DRINKING WATER AND HEALTH one of three reasons: no studies have been conducted in humans with the animal teratogen; the true risk for adverse pregnancy outcome in humans has been underestimated in epidemiological studies; or the animal studies have yielded false-positive results because of test conditions and inter- pretation. The aforementioned FDA literature review indicated that of 165 com- pounds with no evidence of teratogenesis in humans, 29% were negative in all animal species tested and 50% were negative in more than one species. However, 41% of these 165 compounds were positive in more than one animal species. A "nonpositive" response to agents with no evidence of human teratogenesis was observed for 80% of the primates tested, for 70% of the rabbits, for 50% of the rats, for 35% of the mice, and for 35% of the hamsters (Frankos, 19851. Estimates of risk from exposure to individual chemicals are usually based on animal data alone. An understanding of the potential for under- estimating human reproductive risk in epidemiological studies is important in the selection of the safety factor. The appropriateness of using any safety factor in interspecies extrapolation is open to debate; however, factors ranging from 1 to 50 have been found to be adequate for those few agents for which data on minimally effective doses, or lowest- observed-effect levels (LOELs), in humans and animals are comparable (Nisbet and Karch, 19831. That range is based on outcomes measurable at birth and is likely to be inadequate for taking into account early fetal loss in humans. Selection of the no-observed-effect level (NOEL) or the LOEL for use with these safety factors has received more experimental attention. This is discussed in the following section. INTERPRETATION OF ANIMAL DATA When interpreting and extrapolating developmental toxicity data ob- tained from studies in animals, three factors should be considered: the quality and quantity of data available, the relationship between maternal and developmental toxicity, and the selection of the LOEL or NOEL. Quality and Quantity of Data The quality and quantity of the data are the primary factors in deter- mining whether cross-species extrapolation can be performed and the degree of confidence to place in the extrapolation. The quality of data is influenced primarily by the protocol or guidelines under which a study is conducted. An indication of data quality can be obtained by comparing a study protocol with accepted guidelines, such as those developed by the IRLG Testing Standards and Guidelines Work Group (1981) or by the
Developmental Effects of Chemical Contaminants 23 FDA (Collins, 1978; FDA, 19701. In Segment II developmental toxicity studies, pregnant females are exposed during organogenesis and killed one day before estimated delivery, and observations are made on maternal body weight, embryo death, intrauterine growth retardation, and the pres- ence of external, visceral, and skeletal anomalies. Two species are tested, preferably one rodent and one nonrodent. The minimum requirement is three treatment groups and a concurrent control group, each containing 20 rodents or 10 nonrodents (e.g., rabbits). Extrapolation to humans should be made with data on the most sensitive animal species tested, unless there is evidence that this species is not appropriate because of major differences from humans in pharmacokinetics or pharmacodynamics. The quantity of data required is determined largely by statistical power. If an experimental result is found to be statistically significant, i.e., there is a statistically significant association (p < 0.05) at a certain alpha level between exposure and end point, then that study had enough power to detect that difference at the chosen level of significance. If no statistically significant association is found, the study may have had insufficient power to detect a smaller but real effect, or there may have been no effect. Power is related to the number of subjects in a group, to the rarity of the end point, and to the variability in the frequency of the end point's occurrence. In general, the rarer the end point, the fewer the excess occurrences above background needed for detecting an effect. Thalidomide and DES, for example, induced lesions (e.g., phocomelia and vaginal adenocarcinoma) that rarely occur in control populations (e.g., one in 10,000 to 100,0001. Thus, it only took several cases to elevate the inci- dence of these lesions far above the background. Most developmental toxicants tested in animal and human studies cause more common effects, such as intrauterine growth retardation and in- creases in the occurrence of minor anomalies and variants rather than major malformations. In evaluating these end points, it is necessary to consider their background incidence as well as their variance in estimating the sample sizes needed for their detection. Historical control data on the frequency of major and minor fetal mal- formations in conventional laboratory animals are listed in Table 2-3. In developmental toxicity studies, major malformations are considered to be an insensitive indicator of embryotoxicity. Major malformations tend to occur within a relatively narrow range of doses, and unless the compound is a potent teratogen (pattern A in Figure 2-2), large numbers of animals are required to detect a statistically significant increase in their occurrence. With 10 to 20 litters per treatment group, differences of 40% to 50% between treatment and control means are required before statistical sig- nificance is achieved (Palmer, 19811. When more common outcomes, such as embryonic death, intrauterine growth retardation, and elevated
24 DRINKING WATER AND H"LTH TABLE 2-3 Historical Control Data on Major and Minor Fetal Malformationsa Laboratory No. of Fetuses Major Minor Malformations (%) Animal Examined Malformations (%) Visceral Skeletal - New Zealand white rabbits 36,508 0.74 2.53 8.60 CD rats 51,349 0.41 2.02 2.35 CD1 mice 22,389 0.84 3.68 5.32 aAdapted from Palmer, 1978. incidence of common variants and minor anomalies, are taken into con- sideration, a more sensitive appraisal of developmental toxicity can be obtained. The influence on power attributable to the end point's historical vari- ability is illustrated in Table 2-4, which shows the number of litters of different strains of rats and mice that would be required to detect 5% and 10% changes in fetal weight or embryo death. From 22 to 50 litters of mice are required to detect a 10% depression in fetal weight, whereas only 12 to 16 rat litters would be required to detect the same magnitude of weight depression. For embryo death, from 235 to 324 litters of mice are necessary to detect a 10% increase in resorptions, compared to 216 to 248 rat litters. Fewer litters are needed to detect a change in fetal weight, probably because this is a continuously distributed end point with relatively low variability. In contrast, embryo death is a highly variable, binomially distributed parameter, and more than 200 litters are required to detect even a 10% change in this response (Nelson and Holson, 1978~. . TABLE 2-4 Number of Litters Per Group Required to Detect Designated Changes in Fetal Weight and Embryo Lethality in Rats and Micea Change in Change in Fetal Weight Embryo Lethality Test Animal 5% 10% 5% 10% - Mice A/J 84 22 1,176 324 C57BL/6 198 50 992 288 CD1 84 22 805 235 Rats CD 62 16 858 248 Osborne-Mendel 44 12 723 216 aFrom Nelson and Holson, 1978.
Developmental Effects of Chemical Contaminants 25 Given the current testing requirements for 20 rats or mice per group, the most sensitive end point in developmental toxicology studies is fetal body weight, as judged by the statistical power of the study. Within the range of normal variability for this response, a logo change in fetal body weight would be statistically significant (p < 0.05) if that change had been ob- served in 20 rodent litters in a group. For an accurate biological interpretation of depressed fetal weight and embryo death, however, the occurrence of maternal toxicity must be taken into consideration. Most experimental studies of developmental toxicity have been designed to provide information on the basic mechanisms that result in birth defects. Agents are administered under conditions that cause a high incidence of abnormal fetuses, without concern for whether these were secondary to toxic effects in the maternal system. In safety studies, however, failure to recognize that developmental toxicity will inevitably occur at exposure levels causing severe maternal toxicity can lead to false- positive identification of many agents. Relationship between Maternal and Developmental Toxicity Regulatory guidelines for determining the developmental toxicity of chemicals call for dose-response studies in pregnant animals, the highest dose being of sufficient magnitude to induce maternal toxicity (slight but statistically significant weight loss and not more than 10% maternal deaths). The rationale for using a maternally toxic dose is to maximize the potential to detect lesions in the fetus (Palmer, 1981~. Effects observed in offspring at maternally toxic doses are used as a landmark to focus attention on outcomes at lower doses. If a statistically significant incidence of a par- ticular lesion is found in the high-dose group, the biological significance of a lower and perhaps nonsignificant incidence at lower dose levels is magnified. It can be difficult, however, to interpret some effects observed only at maternally toxic dose levels. Are they indicative of unique and selective developmental toxicity, or are they a function of nonspecific alterations in maternal homeostasis? It is generally accepted that developmental toxicity in the form of in- creased resorptions and decreased fetal body weight can occur at mater- nally toxic dose levels. The role of maternal toxicity in the induction of congenital malformations is not clear, however. Recently, Khera (1984) reviewed more than 85 published studies in mice to examine the rela- tionship of birth defects to maternal toxicity and embryo toxicity. He noted that doses of test agents that caused maternal toxicity, as indicated by reduced maternal body weight, clinical signs of toxicity, or deaths, com- monly caused reduction in fetal body weight, increased resorptions, and, rarely, fetal deaths. He identified three patterns of association between
26 DRINKING WATER AND H"LTH maternal toxicity and malformations: (1) for some compounds, maternal toxicity was not associated with malformations; (2) for others, maternal toxicity was associated with a diverse pattern of malformations, which often included cleft palate; and (3) the maternal toxicity of still others was associated with a characteristic and unique pattern of malformations. Compounds in the second category are the most difficult to classify in terms of teratogenic potential. Cleft palate has been reported as the prin- ciple malformation resulting from food and water deprivation during preg- nancy in mice (Szabo and Brent, 19751; however, cleft palate is also a malformation specifically induced in mice by a number of teratogens, most notably the glucocorticoids, without any apparent maternal toxicity. Complete ascertainment of food and water consumption, maternal body weights, and, occasionally, alterations in maternal homeostasis (i.e., organ histopathology, kidney or liver dysfunction, hematological alterations, pharmacologic reactions, and other possible toxic effects) are necessary to distinguish between cleft palate caused by a teratogenic effect of a chemical on the embryo and a nonspecific toxic effect on the dam that secondarily influences embryonic development. Compounds in the third category were structurally unrelated to test agents administered at maternally toxic doses that caused increased re- sorptions and decreased fetal body weight. The characteristic pattern of defects induced by these agents was exencephaly; open eyes; fused, miss- ing, or supernumerary ribs; and fused or scrambled sternebrae. The se- verity and incidence of these defects could be directly related to the degree of maternal toxicity. They were absent or rare at doses that were nontoxic to the dam. Khera (1984) concluded that these defects resulted from maternal toxicity and did not reflect the teratogenic potential of the com- pounds. Kavlock et al. (1985) also examined the association between maternal toxicity and malformations by administering 10 chemicals to mice at doses causing low (Low) to moderate (LD40) maternal mortality rates. Three compounds caused a dose-related increase in the incidence of resorptions, decreased fetal body weight, and malformations that appeared to be in- dicative of developmental toxicity and not the result of indirect maternal action. For seven compounds, an increased incidence of supernumerary ribs was observed. There was a significant (p < 0.001) inverse linear relationship between maternal weight gain during pregnancy and the in- cidence of extra ribs in groups treated with these chemicals, compared to the respective controls. Under the conditions of this study, there appeared to be a quantitative relationship between supernumerary ribs and nonspe- cific maternal toxicity. This quantitative relationship needs to be estab- lished for the defects attributed to maternal toxicity in Khera's study
Developmental Effects of Chemical Contaminants 27 (1984), especially for defects classified as major malformations (e.g., exencephaly and open eyes). As implied from this discussion, not all effects observed in animal studies may be appropriate for use in risk assessment. Although there is little doubt that major effects, such as irreversible and life-threatening malformations and severe embryotoxicity, are deleterious to animals and humans, other effects are of considerable less importance. Common skel- etal variants, such as retarded ossification of the sternum or vertebrae, are considered to be reversible and indicative of slight developmental delay and not of teratogenesis. In humans, the incidence of congenital anomalies of the ribs and vertebrae is low (between 0.02% and 0.03%), and these are considered minor variants with little functional consequence (Heinonen et al., 19771. A low level of concern should be attached to common variations observed in animals, especially if they are the only effects observed and if they only occur in conjunction with maternal toxicity. Greater importance should be given to variations that are dose related or that occur at doses not maternally toxic. Selection of the NOEL/LOEL If the data are of sufficient quality and quantity, it should be possible to identify the NOEL or the LOEL, the maternally toxic dose levels, and the specific types and incidences of adverse effects in the fetus. When the data are insufficient to identify these parameters, the agent should not be subjected to quantitative risk assessment. In the absence of sufficient data, a qualitative assessment should be conducted to rank agents on the basis of high, moderate, or low potential for developmental toxicity in humans. Agents that selectively induce irreversible developmental toxicity in animals, at low doses that are not maternally toxic, have the highest potential for causing developmental toxicity in humans. If maternally toxic exposure causes irreversible developmental toxicity, or if nontoxic ma- ternal exposure causes reversible variants or minor malformations in an- imals, the agent should be considered a moderate hazard to humans. A low hazard to humans can be projected if prolonged exposure of animals to high levels of the compound does not result in developmental toxicity. If the data are sufficient for a quantitative risk assessment, the next decision to be made is whether to establish a NOEL or a LOEL. This decision depends largely on which dose level can most accurately be identified from the data base. Greater experimental confidence can be placed on the LOEL, insofar as this value is empirically derived, whereas the NOEL can be orders of magnitude below the exposure level that would induce developmental toxicity.
28 DRINKING WATER AND HEALTH Selection of the LOEL need not be restricted to responses that are statistically significant. Trends in the data indicating biologically relevant elevations in the incidence of adverse effects at low doses can be used if there are statistically significant increases in the occurrence of these effects at higher doses. LOELs are most accurately selected when the response is minimal (i.e., it is slightly elevated above background and involves reversible developmental toxicity), indicating that the NOEL is being approached. When statistical significance cannot be used as a guide to select the LOEL, which will often be the case, minimal responses can be regarded as those causing a doubling of the background rate (from con- current or historical controls) for the particular response. To protect against the possibility that humans may have double the background rate of the response (which for major malformations would represent an unacceptable increase from approximately 60,000 malformed infants per year to 120,000), a large safety factor can be used for LOELs selected under these conditions. RISK ASSESSMENT Quantitative Assessment of Developmental Toxicity Investigators concerned with the regulatory aspects of risk assessment have focused on the development of a quantitative index for comparing developmental toxicity across species, taking into account concurrent ma- ternal toxicity. Underlying this approach is the perceived need to distin- guish between compounds that are uniquely toxic to the embryo and those that induce developmental toxicity at exposure levels that are also toxic to the mother. Agents in the latter category should be regulated on the basis of their maternal toxicity, whereas those in the former would be regulated on the basis of their unique toxicity to the embryo. Johnson (1980) has developed a testing system for addressing this issue quantitatively. He defined teratogenic hazard potential as the ratio of adult to developmental toxicity, or the A:D ratio, i.e., lowest adult toxic (lethal) dose g lowest developmental toxic dose He has calculated this ratio for more than 70 compounds using data from an in vitro system of Hydra attenuata adult and embryonic tissues. The A:D ratio from the hydra assay has been 0.1 to 10 times greater than the mammalian A:D ratio. Most compounds had ratios near 1, several had ratios larger than 5, and very few had ratios larger than 10 (Johnson and Gabel, 19831. This system has been proposed for setting priorities for further testing of agents in mammalian developmental toxicity studies.
Developmental Effects of Chemical Contaminants 29 Fabro et al. (1982) have explored the quantitative characteristics of a similar type of index in mammalian studies. Dose-response data for adult mortality and fetal malformations were fitted (probit of response against log of dose) for eight compounds. The observed log-probit dose-response lines for lethality and teratogenicity were not parallel, and there was not a constant ratio between the slopes for the two lines. Consequently, a simple ratio between the median lethal dose and the median effective dose (i.e., LD50:EDso) could not be used. To calculate a relative teratogenic index (RTI), Fabro and colleagues established a ratio between one point on each dose-response line. The Loo value was chosen to represent adult mortality on the basis that a low LD value is necessary to guard against compounds with a shallow dose-response curve for adult mortality. The teratogenic dose tDos was chosen for teratogenicity that is, the dose causing a 5% elevation of the malformation rate above background. The investigators believed that the tDos could be estimated with confidence for most teratogens, because induced malformations often occur at a fre- quency between 1% and 20% in animal studies. The committee concluded that this approach appears to be satisfactory for ranking the candidate compounds according to teratogenic potency, provided the relationship of dose to teratogenic response is not complicated by significant adult mor- tality. This ranking system was developed to evaluate structure-teratogenicity relationships between structurally related compounds. For this purpose, the RTI seems adequate. The potential usefulness of this index for inter- species comparisons and risk estimation, however, has not been estab- lished. In their evaluation of the RTI, Hogan and Hoel (1982) argued that because of the lack of parallelism between the probit lines for lethality and teratogenicity, the index will not be invariant in the selection of other LD and tD values; e.g., if a ratio of LD~o:tDos were chosen instead of LDo~:tDos, a different ranking order for the RTI would be obtained. In addition, the index would be subject to the established deficiencies of the probit model, which tends to be insensitive in the low dose region near the origin of the dose-response curve. Therefore, until the RTI has been more extensively applied and eval- uated, it should not be used for risk assessment. It is apparent, however, that a uniform method for ranking agents according to their selective toxicity to the conceptus needs to be established. Such a method would provide a yardstick against which all agents could be compared and would standardize the selection of the NOEL or LOEL for the risk assessment equation. Selection of the safety factor could then be based on the severity of the end point. Existing models for quantitative risk assessment do not appear to be adequate for developmental toxicity data. Multistage models used for
30 DRI N KI NG WATER AND H "LTH mutagenicity and carcinogenicity data are based on a no-threshold as- sumption (Anderson and CAG, 1983), whereas it is generally accepted that thresholds do exist for developmental toxicity (Wilson, 19731. A more thorough discussion of models can be found in Chapter 8. The Environ- mental Protection Agency (EPA, 1983) used a number of models to eval- uate developmental toxicity data on PCBs and found that the safe dose varied by a factor of 7,000 for one set of data, depending on the model used. The EPA and Oak Ridge National Laboratory concluded that existing mathematical models are inappropriate for assessing developmental tox- icity data and that the safety factor approach is appropriate for establishing exposure levels expected to yield acceptable levels of risk (EPA-ORNL, 1982~. The FDA has also indicated that it will use the safety factor approach in developmental toxicity risk assessment but has not given specific details on how the safety factors will be chosen. It is likely that safety factors between 100 and 1,000 will be applied to NOELs identified in develop- mental toxicity studies of drug residues in human food. Smaller factors may be used when the prenatal effect can be ascribed to nonspecific maternal toxicity (Norcross and Settepani, 19831. Thus, due to the absence of other widely accepted approaches, the use of safety factors seems to be the only suitable approach to the quantitative assessment of develop- mental toxicity data (see Chapter 8~. Selection of the Safety Factor The preceding discussion forms the basis of the committee's proposal that the following criteria be considered when selecting a safety factor: · Minimum quality and quantity of data are required to perform a quantitative risk assessment. Compounds without a sufficient data base should be qualitatively assessed for high, moderate, and low potential to cause developmental toxicity in humans. · The committee has concluded that humans should be considered at least 50 times more sensitive than animals to agents causing well-defined developmental toxicity in animal studies. This is in keeping with the fact that safety factors of 100 and 1,000 are most commonly applied to NOELs for developmental end points. · Compounds causing developmental toxicity at levels well below those causing maternal toxicity constitute a greater level of risk than compounds causing developmental toxicity only at maternally toxic doses. This greater degree of risk should be reflected by application of a larger safety factor. · The degree of risk associated with a compound is determined by the severity of response in animal tests and the conditions of time and route
Developmental Effects of Chemical Contaminants 3 ~ of exposure under which the response occurs. The greatest degree of hazard is presented by compounds causing serious effects under conditions of exposure that are relevant to humans. Ultimately, criteria for determining an acceptable risk will have to be developed (Bass and Neubert, 19801. This will involve defining the terms acceptable and risk. What magnitude of a risk is acceptable for a person or for a given population? Is, for example, the doubling of a background rate for an adverse response of 1 in 1,000 acceptable? To what extent should reversible effects and common variants be taken into consideration in risk assessment? These questions cannot be answered from a scientific point of view alone but will require public policy decisions that take into account the benefit of the chemicals under consideration and priorities for protecting public health. Decisions at this level will greatly influence the requirements for safety testing and risk assessment. REFERENCES ACS (American Cancer Society). 1980. Cancer Facts & Figures. National Headquarters, American Cancer Society, Inc., New York. 31 pp. Anderson, E. L., and CAG (Carcinogen Assessment Group of the U.S. Environmental Protection Agency). 1983. Quantitative approaches in use to assess cancer risk. Risk Anal. 3:277-295. Bass, R., and D. Neubert. 1980. Testing for embryotoxicity. Arch. Toxicol. (Suppl. 4):256- 266. Bass, R., D. Oerter, R. Krowke, and H. Spielmann. 1978. Embryonic development and mitochondrial function. III. Inhibition of respiration and ATP generation in rat embryos by thiamphenicol. Teratology 18: 93- 102. Bolande, R. P. 1977. Teratogenesis and oncogenesis. Pp. 293-325 in J. G. Wilson and F. C. Fraser, eds. Handbook of Teratology. Vol. 2. Mechanisms and Pathogenesis. Plenum, New York. Brent, R. L. 1972. Drug testing for teratogenicity: Its implications, limitations and appli- cation to man. Pp. 31-43 in M. A. Klingberg, A. Abramovici, and J. Chemke, eds. Drugs and Fetal Development. Plenum, New York. Brinster, R. L. 1975. Teratogen testing using preimplantation mammalian embryos. Pp. 113-124 in T. H. Shepard, J. R. Miller, and M. Marois, eds. Methods for Detection of Environmental Agents That Produce Congenital Defects. Proceedings of the Guadeloupe Conference Sponsored by l'Institut de la Vie. American Elsevier, New York. Collins, T. F. X. 1978. Reproduction and teratology guidelines: Review of deliberations by the National Toxicology Advisory Committee's Reproduction Panel. J. Environ. Pathol. Toxicol. 2:141-147. Edmonds, L., M. Hatch, L. Holmes, J. Kline, G. Letz, B. Levin, R. Miller, P. Shrout, Z. Stein, D. Warburton, M. Weinstock, R. D. Whorton, and A. Wyrobek. 1981. Report of Panel II: Guidelines for reproductive studies in exposed human populations. Pp. 37- 110 in A. D. Bloom, ed. Guidelines for Studies of Human Populations Exposed to Mutagenic and Reproductive Hazards. Proceedings of conference held January 26-27,
32 DRINKING WATER AND HEALTH 1981, in Washington, D.C. March of Dimes Birth Defects Foundation, White Plains, N.Y. EPA (U.S. Environmental Protection Agency). 1983. Quantitative Risk Assessment of Reproductive Risks Associated with Polychlorinated Biphenyl (PCB) Exposure. Office of Pesticides and Toxic Substances, U.S. Environmental Protection Agency, Washington, D.C. [52 pp.] EPA-ORNL (U.S. Environmental Protection Agency-Oak Ridge National Laboratory). 1982. Pp. 99, 111 in Assessment of Risks to Human Reproduction and to Development of the Human Conceptus from Exposure to Environmental Substances. Report No. EPA- 600/9-82-001. U.S. Environmental Protection Agency, Washington, D.C. Fabro, S., G. Shull, and N. A. Brown. 1982. The relative teratogenic index and teratogenic potency: Proposed components of developmental toxicity risk assessment. Teratogen. Carcinogen. Mutagen. 2:61-76. FDA (U.S. Food and Drug Administration). 1970. Food and Drug Administration Advisory Committee on Protocols for Safety Evaluations: Panel on Reproduction report on repro- duction studies in the safety evaluation of food additives and pesticide residues. Toxicol. Appl. Pharmacol. 16:264-296. FDA (U.S. Food and Drug Administration). 1980. Caffeine; deletion of GRAS status, proposed declaration that no prior sanction exists, and use on an interim basis pending additional study. Fed. Regist. 45:69817-69838. (21 CFR Parts 180 and 182) Frankos, V. H. 1985. FDA perspectives on the use of teratology data for human risk assessment. Fund. Appl. Toxicol. 5:615-625. Fraser, F. C. 1977. Relation of animal studies to the problem in man. Pp. 75-96 in J. G. Wilson and F. C. Fraser, eds. Handbook of Teratology. Vol. 1. General Principles and Etiology. Plenum, New York. Heinonen, O. P., D. Stone, and S. Shapiro. 1977. Pp. 127, 450 in Birth Defects and Drugs in Pregnancy. Publishing Sciences Group, Inc., Littleton, Mass. Herbst, A. L., R. E. Scully, S. J. Robboy, W. R. Welch, and P. Cole. 1977. Abnormal development of the human genital tract following prenatal exposure to diethylstilbestrol. Pp. 399-412 in H. H. Hiatt, J. D. Watson, and J. A. Winsten, eds. Origins of Human Cancer. Book A. Incidence of Cancer in Humans. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Hogan, M. D., and D. G. Hoel. 1982. Extrapolation to man. Pp. 724-727 in A. W. Hayes, ed. Principles and Methods of Toxicology. Raven Press, New York. Hull, D., J. Dobbing, R. W. Miller, F. Naftolin, M. Ounsted, H. Rehder, J. S. Robinson, C. Tudge, and R. H. Usher. 1978. Definition, epidemiology, identification of abnormal fetal growth: Group Report. Pp. 69-83 in F. Naftolin, ed. Abnormal Fetal Growth: Biological Bases and Consequences. Dahlem Konferenzen, Berlin. IRLG (Interagency Regulatory Liaison Group) Epidemiology Work Group. 1981. Guide- lines for documentation of epidemiologic studies. Am. J. Epidemiol. 114:609-613. IRLG (Interagency Regulatory Liaison Group) Testing Standards and Guidelines Work Group. 1981. Recommended Guidelines for Teratogenicity Studies in the Rat, Mouse, Hamster or Rabbit. Washington, D.C. [13 pp.] (Available from National Technical Information Service, Springfield, Va., as Publication No. PB-82-119488.) Ivankovic, S. 1979. Teratogenic and carcinogenic effects of some chemicals during prenatal life in rats, Syrian golden hamsters, and minipigs. Pp. 103-115 in Perinatal Carcino- genesis. National Cancer Institute Monograph 51. NIH Publication No. 80- 1633. National Cancer Institute, U.S. Department of Health, Education, and Welfare, Bethesda, Md. Johnson, E. M. 1980. A subvertebrate system for rapid determination of potential terato- genic hazards. J. Environ. Pathol. Toxicol. 4(5):153-156.
Developmental Effects of Chemical Contam inants 33 Johnson, E. M., and B. E. G. Gabel. 1983. An artificial 'embryo' for detection of abnormal developmental biology. Fund. Appl. Toxicol. 3:243-249. Karnofsky, D. A. 1965. Mechanism of action of certain growth-inhibiting drugs. Pp. 185- 194 in J. G. Wilson and J. Warkany, eds. Teratology: Principles and Techniques. University of Chicago Press, Chicago. Kavlock, R. J., N. Chernoff, and E. H. Rogers. 1985. The effect of acute maternal toxicity on fetal development in the mouse. Teratogen. Carcinogen. Mutagen. 5:3-13. Khera, K. S. 1984. Maternal toxicity A possible factor in fetal malformations in mice. Teratology 29:411-416. Kimmel, C. A., J. F. Holson, C. J. Hogue, and G. L. Carlo. 1984. Reliability of Exper- imental Studies for Predicting Hazards to Human Development. Final Report. NCTR Technical Report for Experiment No. 6015. National Center for Toxicological Research, Jefferson, Ark. 56 pp. Kline, J., Z. Stein, B. Strobino, M. Susser, and D. Warburton. 1977. Surveillance of spontaneous abortions: Power in environmental monitoring. Am. J. Epidemiol. 106:345- 350. Miller, R. W. 1973. Prenatal origins of cancer in man: Epidemiological evidence. P. 175 in L. Tomatis and U. Mohr, eds. Transplacental Carcinogenesis. IARC Scientific Pub- lications No. 4. International Agency for Research on Cancer, Lyon, France. Nelson, C. J., and J. F. Holson. 1978. Statistical analysis of teratologic data: Problems and advancements. J. Environ. Pathol. Toxicol. 2:187-199. Neubert, D., H. J. Barrach, and H.-J. Merker. 1980. Drug-induced damage to the embryo or fetus: Molecular and multilateral approach to prenatal toxicology. Curr. Top. Pathol. 69:241-331. Nisbet, I. C. T., and N. J. Karch, eds. 1983. Chemical Hazards to Human Reproduction. Noyes Data Corporation, Park Ridge, N.J. 245 pp. Nishimura, H., and T. Tanimura. 1976. Clinical Aspects of the Teratogenicity of Drugs. American Elsevier, New York. 453 pp. Norcross, M. A., and J. A. Settepani. 1983. Current guides to establishing safe levels of animal drug residues. Paper presented at the International Symposium on the Safety Evaluation of Animal Drug Residues, Berlin, Federal Republic of Germany, 27-29 September 1983. Ornoy, A., J. Salamon-Arnon, Z. Ben-Zur, and G. Kohn. 1981. Placental findings in spontaneous abortions and stillbirths. Teratology 24:243-252. Palmer, A. K. 1978. The design of subprimate animal studies. Pp. 215-253 in J. G. Wilson and F. C. Fraser, eds. Handbook of Teratology. Vol. 4. Research Procedures and Data Analysis. Plenum, New York. Palmer, A. K. 1981. Regulatory requirements for reproductive toxicology: Theory and practice. Pp. 259-287 in C. A. Kimmel and J. Buelke-Sam, eds. Developmental Tox- icology. Raven Press, New York. Pratt, R. M., and D. S. Salomon. 1981. Biochemical basis for the teratogenic effects of glucocorticoids. Pp. 179-199 in M. R. Juchau, ed. The Biochemical Basis of Chemical Teratogenesis. Elsevier/North-Holland, New York. Ritter, E. J. 1977. Altered biosynthesis. Pp. 99-116 in J. G. Wilson and F. C. Fraser, eds. Handbook of Teratology. Vol. 2. Mechanisms and Pathogenesis. Plenum, New York. Schardein, J. L. 1976. Drugs as Teratogens. CRC Press, Inc., Cleveland. 291 pp. Shepard, T. H. 1980. Catalog of Teratogenic Agents, 3rd ed. The Johns Hopkins University Press, Baltimore, Md. 410 pp.
34 DRINKING WATER AND H"LTH Staples, R. E. 1975. Definition of teratogenesis and teratogens. Pp. 25-26 in T. H. Shepard, J. R. Miller, and M. Marois, eds. Methods for Detection of Environmental Agents That Produce Congenital Defects. American Elsevier, New York. Strobino, B. R., J. Kline, and F. Stein. 1978. Chemical and physical exposure of parents: Effects on human reproduction and offspring. J. Early Hum. Dev. 1:371-399. Szabo, K. T., and R. L. Brent. 1975. Reduction of drug-induced cleft palate in mice. Lancet 1: 1296-1297. Wilson, J. G. 1965. Embryological considerations in teratology. Pp. 251-261 in J. G. Wilson and J. Warkany, eds. Teratology: Principles and Techniques. University of Chicago Press, Chicago. Wilson, J. G. 1973. Environment and Birth Defects. Academic Press, New York. 305 pp.
