Tobacco Documents Online

Reproductive Toxicology, Vol. 3, pp. 3-12, 1989 - 0890-6238/89 $3.00 + .00 Printed in the U.S.A. Copyright © 1989 Pergamon Press plc • Special Article CRITERIA FOR IDENTIFYING AND LISTING SUBSTANCES KNOWN TO CAUSE DEVELOPMENTAL TOXICITY UNDER CALIFORNIA'S PROPOSITION 65* DONALD R. MATTISON Division of Human Risk Assessment, National Center for Toxicological Research, Jefferson, Arkansas 72079 and Division of Reproductive Pharmacology and Toxicology, Department of Obstetrics and Gynecology, Slot 518, University of Arkansas for Medical Sciences, 4301 Markham, Little Rock, Arkansas 72205 JAMES W. HANSON Department of Pediatrics, College of Medicine, University of Iowa, Iowa City, Iowa 52242 D.M. KOCHHAR Department of Anatomy, Jefferson Medical College, Philadelphia, Pennsylvania 19107 K.S. RAO Toxicology Research Laboratory, The Dow Chemical Company, Midland, Michigan 48674 L BACKGROUND The International Life Sciences Institute-Nutrition Foundation (ILSI-NF) convened a group of experts in developmental toxicity to discuss criteria for listing substances as developmental toxicants under the pro- visions of California's Safe Drinking Water and Tox- ics Enforcement Act of 1986 ("Proposition 65") (1). The Expert Committee on Criteria for Listing Devel- opmental Toxicants developed this document with technical coordination provided by the ILSI Risk Sci- ence Institute to provide additional guidance to the Governor's Scientific Advisory Panel (SAP) as it con- siders criteria for listing substances "known to cause developmental toxicity." In addition, this document was formally peer reviewed by eight other experts in the area of developmental toxicology. The Expert Committee reviewed the "Guidelines (Criteria) for Listing Chemicals as Reproductive Tox- icants" (2) proposed by the Reproductive Toxicity Subpanel of the SAP and found there was considerable agreement among Expert Committee members with the principles outlined in the Subpanel's document. In addition to the Subpanel's document, the Expert Committee also reviewed other guidelines and crite- ria, most notably the U.S. Environmental Protection Agency guidelines for developmental toxicants (3). Thus, the Expert Committee, built on existing criteria and made modifications deemed necessary to meet the requirements and objectives of Proposition 65. The Expert Committee focused its efforts on establishing criteria specific to Proposition 65 and only for developmental toxicity endpoints. This focus was recommended by scientists who were contacted to participate in the project. The Expert Committee would like to point out that the approach taken in Proposition 65 combines the risk assessment and risk management processes into one step in the listing process. This is a major dif- ference in approach and must be kept in mind when comparing other processes used by federal and other state regulatory agencies. 'The authors are members of The Expert Committee on Criteria for Listing Developmental Toxicants convened by the International Life Sciences Institute-Nutrition Foundation. The Expert Committee wishes to extend its appreciation to the ILSI-NF Proposition 65 Task Force, Dr. Catherine St. Hilaire, ENVIRON Corporation, and the ILSI Risk Science Institute staff for their assistance in the preparation of this document, in particular Dr. Carol J. Henry, Ms. Sarah Connick, Dr. Irene B. Glowinski, Ms. Stephanie D. Carter, and Ms. Gretchen Bretsch. Correspondence and reprint requests should be directed to Dr. Mattison at the University of Arkansas for Medical Sciences. 3

4 Reproductive Toxicology II. OBSERVATIONS ON PROPOSITION 65 AND APPROACH TO DEVELOPING CRITERIA FOR LISTING In the course of its deliberations, the Expert Committee made the following observations: 1) Proposition 65 is clearly a' law intended to impact on and to improve the public health and well- being of the citizens of California. 2) The requirement that substances be shown to have no observable effect assuming exposure at 1000 times the level in question is a component of the law and, as such, cannot be considered in the listing pro- cess. This requirement does not permit flexibility and scientific judgment in the determination of acceptable exposure levels for listed substances. 3) Proposition 65 places great importance on the listing process, especially in the case of reproductive and developmental toxicants. When such substances are listed, further consideration of most risk assess- ment, risk management, and risk communication con- cerns is highly constrained so that listing is the primary means for insuring sound consideration of the ulti- mate public health impact of a given substance. Further discussion of each of the above conclu- sions of the Expert Committee is presented below. A. Public health considerations Proposition 65 was adopted by California voters to protect themselves against and to be informed about chemicals that cause cancer, birth defects, or other reproductive harm. Viewed from a public health perspective, the state has a clear obligation to protect and promote the public health in accord with these desires. The appropriate implementation of a program to address these needs and concerns, however, goes beyond risk assessment and communication. Further- more, different parties to this process may understand these needs and concerns from a variety of perspec- tives. Risk assessment and interpretation of scientific data occur in a complex context, and public percep- tions may not be in accord with scientific interpreta- tion of available data. Promotion of the public health includes not only addressing the need for accurate information on risks stemming from exposure, but also the facilitation of health care and appropriate responses from both ex- posed persons and health-care providers. From this perspective it is insufficient to provide information on possible hazards while failing to ;provide a means of management of concerns engendered. To do so would be dysfunctional and disobeys the first principle of health care: "Above All Do No Harm." It is equally important not to falsely reassure or falsely frighten. Volume 3, Number 1, 1989 An adequate public health response and policy must also address access to health care, quality, and cost of health care, as well as equality, ethical, and legal issues. Furthermore, the active participation of all par- ties in the continuum of risk assessment, risk com- munication, and risk management must be assured. In the case of developmental toxicity, these parties include not only consumers (mothers, fathers, and fetuses and infants) but also health-care providers and institutions, scientists, state agencies, information media, and the private economic sector. The Expert Committee would like to stress that although the need for adequate education of the public and of health- care practitioners is important for all toxic endpoints addressed by Proposition 65, it is especially critical for potential developmental toxicity outcomes since a pregnant woman learning after the fact that exposure has occurred to a "known" developmental toxicant may have various options concerning actions that can be taken (amniocentesis, counseling, terminating the pregnancy, etc.). Education and effective communication programs are essential in order to avoid the untoward public health effects of warning. Obviously, warning of pos- sible hazard of exposure may lead individuals to de- cide whether or not to limit or avoid exposure based on an individual assessment of personal risks and benefits of the exposure. To allow individual discretion in decision making, access to the data and to appro- priately trained, knowledgeable health-care practi- tioners is essential. The Expert Committee suggests that the state Health and Welfare Agency actively develop education plans for the public and health care practitioners that will precede and continue through- out the institution of warning requirements. The state should consider developing guidelines that suggest management strategies to assist health care profes- sionals in dealing with these issues. The Expert Committee also recommends that, when listing compounds, the SAP explicitly identify the type or types of toxicity caused (male, female and/or developmental toxicity or cancer) and that the SAP request that this information be included in the official Governor's list. Also, the Expert Committee recommends the Scientific Advisory Panel indicate, if known, the site and mechanism of action, and the critical developmental periods during which toxicity may occur. B. Safety factors In defining "known to cause developmental tox- icity," the Expert Committee did not consider the impact of the 1000-fold factor incorporated in the law

Criteria for developmental toxicity • D. R. MATTISON ET AL. as a mandatory "uncertainty factor" for establishing acceptable human levels from the experimental no- observed-effect levels (NOEL).' The Expert Commit- tee, however, felt that it must comment on the lack of scientific basis for this approach to establishing ac- ceptable exposure levels for developmental (or repro- ductive) toxicants. This approach is not scientifically defensible. No single uncertainty factor is appropriate for all situations. The data available to determine that a substance is "known •to cause developmental toxic- ity" will vary tremendously in quantity, quality, and type. In addition, the kinds of effects caused, the sen- sitivity of the conceptus in the species studied, the nature of the dose-response relationship, and the site and mechanism of action should influence the deter- mination of the levels that are acceptable for human exposure. Although the Expert Committee acknowl- edges that the "correctness" of the uncertainty-factor approach cannot be objectively assessed, this approach has been adopted by all regulatory and advisory groups that have addressed developmental toxicity risk assess- ment, including the U.S. Environmental Protection Agency (EPA), U.S. Food and Drug Administration, World Health Organization, and the National Acad- emy of Sciences (NAS). The appropriate magnitude of an uncertainty factor is influenced by many of the factors described above, all of which should be con- sidered. In this regard, the Expert Committee believes that an approach such as that taken by the EPA in its guidelines for developmental toxicity risk assessment (3) is preferable to an across-the-board imposition of a "standard factor" of any magnitude: The uncertainty factor approach results in a calculated exposure level believed to be unlikely to cause any toxic developmental re- sponse in humans. The size of the uncertainty factor will vary from agent to agent and will require the exercise of scientific judgment, taking into account interspecies differences, the nature and extent of human exposure, the slope of the dose-response curve, the types of developmental effects observed, and the relative dose levels for maternal and developmental toxicity in the test species. The Expert Committee recommends an approach that would permit the selection of an appropriate uncertainty factor for a listed substance based on the above principles. This flexibility also would permit and encourage the development of newer, more objec- tive means of deriving acceptable human exposure levels based on, for example, physiologically based pharmacokinetic approaches or models that improve the confidence in cross-species extrapolation. 'The NOEL describes an exposure level at which no effect of any type is observed in the test popula`tion. The term "no-observ- able-adverse-effect-level" (NOAEL) describes an exposure level at which there are no biologically relevant effects in the test popula- tion. In the case of many exposure levels, the NOEL and NOAEL both refer to the same exposure levels. 5 C. Listing as it relates to risk assessment, risk management, and risk communication As the law currently stands, the listing process automatically initiates a prohibition of discharge of the listed substance into sources of drinking water and requires that consumers be warned that products con- tain a substance or substances "known to the State of California to cause reproductive toxicity" if the sub- stance is present in the discharge or in the product at a level that exceeds 1/ 1000 times the NOAEL. With the exception of selecting an appropriate NOAEL, no further scientific evaluation of the risk or regulatory review is required following listing. Thus, the listing process substitutes for all aspects of risk assessment, risk management, and risk communication, except for the identification of the NOAEL. (In practice, the NOAEL has been addressed by the SAP when sub- stances such as alcohol were recommended for listing, and thus its consideration could be considered as a component of the listing process.) For this reason, the Expert Committee has developed criteria for listing developmental toxicants that include many compo- nents of the dynamic process of risk assessment. The Expert Committee would like to point out that the approach taken in Proposition 65 of combin- ing elements of risk management and aspects of risk assessment into a single step is a significant departure from the approaches taken by federal and state agen- cies. The usual approach taken was described in the NAS study of risk assessment (4) in which risk assess- ment is deliberately separated from risk management. In addition, the risk assessment itself is divided into four distinct steps. Given that all four steps of risk assessment as well as risk management are embodied in the listing process for reproduction and develop- mental toxicants, the Expert Committee developed listing criteria that emphasize the critical factors in risk assessment: hazard identification, dose-response assessment, human exposure assessment, and risk char- acterization. The Expert Committee also made addi- tional recommendations to the SAP and appropriate state officials, which were highlighted in the preced- ing sections, aimed at the risk management outcome of improving the public health through the listing process. III. CRITERIA FOR LISTING DEVELOPMENTAL TOXICANTS A. Preamble As in all categorization methods developed to aid in the evaluation of scientific evidence relied upon in risk assessment, there is an inherent danger of such methods becoming oversimplified, "cookbook" ap-

6 Reproductive Toxicology proaches that are viewed as inflexible and result in inappropriate constraints on scientific judgment. The Expert Committee stresses that the principles delineated in this document have been constructed to require, throughout the process, the exercise of professional expert judgment of the significance of each piece of evidence and the overall significance of the evidentiary picture that emerges from a weight-of-evidence deter- mination. Thus, each listing decision made by the SAP will be guided by the general principles outlined in this document but will be, in the final analysis, based on a professional judgment of the overall likeli- hood that a given substance will be a developmental toxicant in humans. Of the endpoints addressed by Proposition 65, the Expert Committee firmly believes developmental toxicity is especially problematic in that a warning of possible fetal harm due to exposure often occurs be- fore pregnancy is either suspected or confirmed, thus presenting individuals so exposed with a choice of whether or not to continue the pregnancy to term. Clearly, a "cookbook" approach that does not require full scientific consideration could create a public health dilemma unparalleled in the history of measures such as Proposition 65, which are aimed at improving the public health. Obviously, the SAP (and the Governor, who makes the final listing decisions) must constantly bal- ance the desire to prevent exposures to substances having sufficient evidence to justify concern with the knowledge that notice of possible adverse fetal effects received "after the fact" could also result in inappro- priate and untoward public responses. The involve- ment of clinicians in this decision-making process is recommended. The Expert Committee would like to emphasize again the discussion in this document's background section of the essential role of educating both the professional health care providers/public health practitioners and the general public about the meaning of warnings of exposure to substances "known to cause developmental toxicity." Finally, risk assessment guidelines and criteria for making weight-of-evidence determinations are bound by the context within which they were developed (in this case, the requirements of Proposition 65) and the point in time at which they were developed. That is, the criteria presented may or may not be useful within other public health/regulatory contexts; thus, they must be fully evaluated before adopting them into other contexts. Further, the Criteria may or may not be valid as new scientific understanding is gained in the areas of developmental toxicology and risk assess- ment methodology. These criteria, or any other set of criteria adopted for use under proposition 65, must be reviewed and updated periodically. Volume 3, Number 1, 1989 B. Approach to development of criteria The Expert Committee adopted the following definition of developmental toxicity: adverse effects on the developing organism that may result from exposure prior to conception (either parent), during prenatal development, or post-natal to the time of sexual maturation. In developing the definition of "known to cause developmental toxicity" and the principles and crite- ria for listing substances, the Expert Committee relied upon a "weight-of-evidence" approach that encour- ages the evaluation of biological plausibility and hu- man relevance. Making determinations of potential human haz- ard, such as the determination that a substance is "known to cause developmental toxicity," requires that all information available on that substance be consid- ered. All available appropriately conducted studies (animal and human) should be evaluated to reach an overall determination based on the "preponderance" or "weight" of the overall evidence. Most chemicals are capable of interfering with in utero development if exposure is sufficiently high; therefore, a major consideration must be the dose or level of exposure, whether in experimental animals or humans. Other types of information can provide addi- tional insights for this evaluation. For example, a structural similarity to known developmental toxi- cant(s) may strengthen the association between the substance under consideration and developmental tox- icity. Conversely, evidence of an adverse developmen- tal effect in a single species or multiple species that are known to metabolize and distribute the substance differently from humans Would be weighted less. In addition, because genetic heterogeneity is known to be important in individual human susceptibility to toxi- cants (5,6) and because considerable interspecies and interstrain variation in susceptibility exists among ani- mal species and strains, results from animal models of similar sensitivity to humans would be weighted more heavily. Data used in an overall weight-of-evidence deter- mination include those derived from: • epidemiologic studies, • human case studies, • animal toxicity studies (including reproduction studies), • pharmacokinetic studies, • pharmacogenetic studies, • in vitro studies, and • structure-activity analyses. The Expert Committee emphasizes that a well- conducted study (animal or human) showing no ad- verse effect on development can provide valuable information for a weight-of-evidence determination.

9 Criteria for developmental toxicity e D. R. MATTISON ET AL. 7 Because negative studies are reported less frequently in the literature, the Expert Committee recommends that the SAP encourage early submission (as soon as possible) of all pertinent information, including un- published negative studies having sufficient documen- tation, so that an informed scientific judgment can be made regarding the results. In the discussion of criteria for identifying sub- stances "known to cause developmental toxicity," weight-of-evidence determinations for human data, animal data, and combined human and animal data are discussed. The Expert Committee would like to emphasize that although similar terms may be used for developmental endpoints, they often have differ- ent interpretations and implications for animal and human investigations. In evaluating the degree of risk to the human conceptus, a number of differences that exist among animal species and between animals and humans must be carefully considered. These include genetic, metabolic, anatomic, physiologic, and other interspecies differences in development. Differences in the anatomy and physiology of the placenta between experimental animals and humans also need to be considered. These factors can influence the results of developmental toxicity testing. C. Definition of "known to cause developmental toxicity" The Expert Committee developed the following definition of a developmental toxicant: A substance is "known to cause developmental toxicity" within the context of Proposition 65 if there is a consistent pattern of adverse effect(s) in humans or a predictable pattern of adverse effect(s) in animals combined with human relevance and biological plausibil- ity, indicating that human exposure under the appropriate condi- tions (of sufficient magnitude at susceptible times) is likely to result in developmental toxicity. D. Recommended criteria for identifying and listing substances "known to cause developmental toxicity" A substance is identified as "known to cause developmental toxicity" if it meets the requirements in either Category I or 2: Category 1. Sufficient evidence in humans. Suffi- cient human evidence for listing of a substance as being "known to cause developmental toxicity" must consist of a) convincing evidence of an association between an exposure to a substance and a consistent pattern of abnormal outcomes; and b) biological plausibility of a causal relationship. Failure of the data to meet either of these crite- ria signals a need for further' investigation before a causal relationship is firmly asserted. A detailed dis- cussion of issues related to evaluation of human evi- dence is found in Part IV. (a) Convincing evidence of an association be- tween an exposure to a substance and a consistent pattern of abnormal outcomes. Such evidence implies the availability of studies from which the weight-of- evidence supports and establishes the substance as a crucial factor within a chain of causal events. The data from such studies should be reproducible, valid, sufficiently sensitive and specific, and should provide evidence clearly linking exposure to individuals with known outcome. Evidence of a dose-response rela- tionship would further support a causal relationship. The most persuasive evidence for a cause-effect relationship arises when a number of studies con- ducted by different investigators at different times using different methodologies in different geographic or cultural settings with different populations all show similar results. In some cases, data from several stud- ies can be pooled for analysis to test a hypothesis. (b) Biological plausibility of a causal relationship. In forging the chain of causality, arguments of biolog- ical relevance and plausibility must be kept constantly in mind. Unexplained inconsistencies require caution in interpretation even of well-designed and powerful studies that identify statistically significant associa- tions. Statistical significance does not confer biologic validity. The Expert Committee believes that if a gen- eral hypothesis for causation cannot be developed, sufficient evidence in humans may not have been obtained. Category 2. Sufficient evidence in animals. Suffi- cient animal evidence for listing of an agent as being "known to cause developmental toxicity" must consist of a) Consistent pattern of adverse developmental outcomes; b) Demonstrated dose-response relationship; and c) Biological plausibility. Failure of the data to meet all of these criteria signals a need for further investigation before it can be established that sufficient evidence exists. A de- tailed discussion of issues related to evaluation of ani- mal evidence is found in Part IV. (a) Consistent pattern of adverse developmental outcomes. The quality and quantity of animal data vary considerably among tested substances. Often, a major consideration is whether there are adequate data to evaluate the substance. In general, the ideal data set should include results from developmental toxicity studies in two species, preferably rodent and nonrodent. Treatment should be via the likely route of human exposure. The highest dose should be the one that causes measurable maternal toxicity. Results should permit identification of the NOAEL and low-

8 Reproductive Toxicology est-observable-adverse-effect level (LOAEL) for ma- ternal or developmental toxicity. From a given data set, a "consistent pattern of outcomes" would consist of adverse developmental effects in multiple species tested, or adverse effects in multiple strains, or multiple studies of the same strain of a single species. The types of developmental effects seen in different species or strains need not be the same to satisfy this requirement. In making a judg- ment concerning consistency, multiple well-conducted studies showing an absence of adverse effects on development, even in the presence of studies that demonstrate developmental toxicity, indicate the need to be especially careful in evaluating the biologic plau- sibility and human relevancy of the observation of adverse effects. In all these determinations, differ- ences in dosing and/or exposure regimen need to be factored into a finding of "consistent" results. The Expert Committee has refrained from indi- cating an "appropriate" number of studies showing adverse effects that is needed for listing because there is no standard formula for balancing the number of studies showing adverse effects versus the number of studies demonstrating an absence of effects. Also, one could envision a situation where unequivocal evidence of an adverse developmental effect is gained in a sin- gle study in which the substance is administered via a likely route of human exposure. In that case, the SAP could judge that the substance should be listed. (b) Demonstrated dose-response relationship. Confidence in the results of individual studies is greatly increased if a clear relationship between increasing dose and increasing response (as measured by increased incidence or severity of effect) is demonstrated. In the evaluation of developmental studies per- formed in animals, consideration should be given to excluding maternal toxicity as a confounding factor. Adverse effects on development that occur only when maternal toxicity occurs may not indicate a specific or unique hazard to the conceptus. In contrast, develop- mental toxicity in the absence of maternal toxicity may imply unique susceptibility of the conceptus. . In any study, calculation of the ratio of the larg- est dose that produces no adverse effect to the mother to the largest dose that produces no adverse effect to the conceptus can be used to assess the extent of dif- ferential susceptibility. A large ratio (greater than 2) means that developmental toxicity occurs at doses far lower than those producing toxicity to the mother. A small ratio (approximately 1) implies that the dose exerting toxicity in the conceptus may be close to the dose producing adult toxicity. It is important to reach a judgment concerning the likelihood that develop- mental toxicity would occur in the absence of mater- nal toxicity. The Expert Committee recommends that Volume 3, Number 1, 1989 every effort be made to reach a judgment in this re- gard and that this judgment be factored into the list- ing process. Different methods for evaluating the relationship between maternal and developmental tox- icity have been suggested (7-9); thus, different ratios can be derived. There is no general rule concerning the selection of a ratio that is the definitive value for determining selective developmental toxicity. The Ex- pert Committee recommends that the range of these methods be considered when evaluating the develop- mental hazard of a particular substance. An approach to applying this criterion would be to examine all data available from animal studies, to determine the NOAELs for maternal and develop- mental toxicity, and to consider the human exposure potential for that substance and its relationship to the NOAELs. Substances having relatively high maternal- to-conceptus toxicity ratios are more likely to be de- velopmental toxicants in humans exposed to environ- mental levels of the substance that are well below the maternal toxic dose. In some cases, human exposure is near the maternally toxic dose. Substances that show developmental effects at doses associated with mater- nal toxicity with human exposure at or near the ma- ternally toxic dose should be considered for listing. .(c) Biological plausibility. Data that are consis- tent with a hypothesis (even a general one) concerning the possible mechanism of action for a substance are sufficient to satisfy this criterion. Biological plausibil- ity is strengthened by evidence from in vitro studies; by structural similarity of a substance to known devel- opmental toxicants; and by knowledge of uptake, metabolism, distributions, etc., that supports a possi- ble developmental effect. In summary, if the above three criteria are met, human relevance is assumed. In addition, positive re- sults in a species that metabolizes, distributes, and transfers a given chemical across the placenta in a manner similar to humans strengthens confidence that the animal findings are relevant to humans. IV. SCIENTIFIC JUDGMENT REQUIRED FOR EVALUATION OF SUBSTANCES FOR SUFFICIENT EVIDENCE OF DEVELOPMENTAL TOXICITY A. Sufficient evidence in humans Criteria for evidence of human teratogenicity (a subset of developmental toxicity) have been described in previous publications (10=13). 1. Developmental endpoints in humans. Human outcomes of interest as measures of developmental toxicity include alterations of growth, structure, and function, in addition to death (Table 1). It is impor- tant, however, to recognize that these may not be in-

Criteria for developmental toxicity • D. R. MATTISON ET AL. 9 dependent events. That is, for many, if not most, truly hazardous environmental exposures there is a spec- trum of outcomes that may vary in frequency, se- verity, and type. It is uncommon among an exposed population to find individuals who display all the im- portant structural and functional consequences attrib- utable to that exposure. In other words, variability of outcomes is the rule. The sources of this variability are well known: differences in dose, timing of expo- sure, host susceptibility (both maternal and fetal), and interactions with other environmental factors. It is also important to note that structural defects as an outcome of hazardous environmental exposures occur in characteristic patterns, not as random aggre- gates of defects. Individual defect categories (espe- cially when classified by organ system or body region) are both etiologically and pathogenetically heteroge- neous. Furthermore, many of the adverse outcomes listed in Table 1 are measured and classified in differ- ent ways, and in some cases in greater detail, than in animal experiments. This is especially true for birth defects, growth disturbances, and abnormalities of function, such as learning and behavior. The classification of human structural abnormal- ities is different from that for animal abnormalities. Table 1. Examples of human endpoints Human structural abnormalities include malforma- tions, disruptions, and deformations (14). These types of abnormalities have different pathogenetic and etio- logic implications than those for animals. Further- more, some structural defects in humans may be considered to be normal variations or mild abnormal variations of no clinical significance while, nonethe- less, being important clues to mechanisms of abnor- mal development (15). Thus, there are many different types of endpoints or outcomes in humans that are identified as develop- mental effects. These endpoints are measured differ- ently and interpreted differently from those observed in animal studies. The Expert Committee emphasizes the importance of involving skilled laboratory scien- tists and clinicians in the conduct and interpretation of human studies. This is especially true for studies in which postnatal functional abnormalities (e.g., learn- ing and behavior) are endpoints of concern. As for other types of toxic endpoints, a single finding of al- tered function that is not correlated with other related findings is of questionable significance and should be considered cautiously in making a determination of developmental toxicity. 2. Study design and interpretation issues. Data that link exposures of human populations to specific adverse developmental outcomes and that can be re- lied upon to reach a determination of "sufficient evi- dence" may be accumulated from two sources: clinical of developmental toxicity studies and epidemiological investigations. The ad- Fetal death (early and late) Stillbirth Perinatal death Placental, cord, and fetal membrane abnormalities Infancy and childhood mortality Infancy and childhood morbidity Intrauterine growth retardation Postnatal growth retardation Proportionate, disproportionate Symmetrical, asymmetrical System limited, generalized Change in gestational age at delivery Altered sex ratio Birth defects Major, minor, mild Malformations, deformations, disruptions Single defects, syndromes, sequences, patterns Mutations, chromosomal defects, monogenic disorders Abnormal maturation Abnormal sexual development or function Mental retardation/learning disability Specific organ system dysfunction Development disabilities Visual impairment Hearing impairment Cerebral palsy and other motor handicaps Other sensory disturbances Behavioral disorders Transplacental carcinogenesis and mutagenesis (genotoxicity)* *This endpoint raises concerns that are not specifically addressed in this document. vantages and disadvantages of each have been thor- oughly discussed in the literature. It is clear, however, that these approaches to human risk assessment are complementary. Clinical investigations, including case reports, are especially valuable in formulating hypotheses, and on occasion, offer unique insights into pathogenetic mechanisms and clues to etiology. The relevance of results from animal models to human experience is best determined through comparison to human struc- ture and function. In fact humans are among the best (and most easily) studied organisms in many respects, and the insights gained from astute clinical observa- tions should not be overlooked. Most agents recog- nized as known human developmental toxicants were first identified by alert clinicians (16). Epidemiologic investigations of human popula- tions may employ a variety of study designs. Each has unique advantages and disadvantages that must be considered when evaluating the results. Particularly important issues to weigh are the potential sources of bias (e.g., ascertainment problems) and confounding factors, as well as the sensitivity and specificity of a given design. The size of the study population should offer sufficient statistical power to detect the relevant

' 10 Reproductive Toxicology outcomes. Sources, types, and quality (including reli- ability) of data need to be carefully examined, as does the choice of control (comparison) populations. When there is evidence of differences in susceptibility in human populations, this fact should be considered in the design of epidemiologic and clinical studies (17). Detailed discussion of the strengths, weaknesses, and study design considerations for epidemiologic and clinical studies have been published elsewhere (18) and are not reviewed in detail here. Most developmental toxicants produce their ef- fects during specific critical developmental periods, which vary across compounds and species. When pos- sible, human epidemiologic studies should demon- strate that exposure of the study population occurred at the times when the individuals were susceptible. Since no single developmental toxicant increases risk uniformly for all effects, it is essential to establish strict diagnostic criteria for endpoints of interest. Exposure to a developmental toxicant may result in multiple adverse effects (lethality, altered growth, malformations, and functional impairment). Thus, studies must be carefully designed because a suspect agent may result in a spectrum of abnormalities fol- lowing exposures at different times during pregnancy. The choice of statistical procedures to be applied to particular study designs can sometimes influence interpretation, especially if the procedures are based upon inappropriate biologic models or theories. Sim- ple statements of point estimates of relative risk or odds ratios are meaningless without concomitant cal- culations of confidence intervals.: Furthermore, cal- culations of statistical significance must always be considered in light of the number of comparisons made and the biological plausibility of associations noted. B. Sufficient evidence in animals 1. Developmental endpoints in animals. End- points in developmental toxicity can include growth retardation, death of the conceptus, deleterious struc- tural malformation (teratogenesis), and functional def- icits. Table 2 presents some examples of developmental endpoints observed in laboratory animals. A structural abnormality can be classified as a malformation or variation. A malformation (terato- genicity) is usually defined as a permanent structural change that may adversely affect survival, develop- ment, or function. A variation is a divergence beyond the usual range of structural constitution that may not adversely affect survival or health. Altered growth is an alteration in offspring organ br body size and, in some instances, is a reversible phenomena. Functional deficiency is a delay or deficit in functional compe- tence of the organism or organ system. Delays in the attainment of functional competence may be revers- Volume 3, Number 1, 1989 Table 2. Examples of developmental alterations observed in laboratory animals Major effects Cleft lip/palate Aphakia Anophthalmia Renal agenesis Malformed heart valves, vessels Gastroschisis Missing ribs, vertebrae Exencephaly Spina bifida Malformed or missing limbs Fetal death Increased number of resorptions Variations Delayed ossification of bones Lumbar ribs Wavy ribs Unfused centers of ossification Extra center of sternebral ossification Increased renal pelvic cavitation Hemorrhages at some sites Some displaced testes Some types of hydroureter Adapted from Wang and Schwetz (19). ible. In general, reversible effects that occur in the presence of maternal toxicity are judged to be of less concern in the determination of potential human risks. The developmental toxicity endpoints encoun- tered in experimental animals do not and should not necessarily be expected to mimic those observed in humans exposed to the same toxicant. Similarly, spe- cific substance-related endpoints in humans are not always reproduced in experimental animals. The ab- sence of absolute uniformity qf response is not sur- prising, however, when the many critical differences that exist between the conditions of human and animal model exposure are considered. For example, differ- ences in dosage, placentation, metabolism, pharmaco- kinetics, critical periods of development, durations of gestation, etc., can be expected to affect expression of developmental toxicity. Because the present understanding of functional and behavioral toxicity as derived from conventional toxicology studies is sometimes compromised by defi- ciencies in study protocols, data collection methods, and identification of appropriate endpoints, the Ex- pert Committee recommends that data based on these endpoints be considered with caution. The relevance of findings of behavioral and other functional deficits is strengthened when data from neurologic, neuro- physiologic, neuropathologic, or neurochemical stud- ies are available to evaluate the significance of the functional deficits data (20). 2. Statistical and interpretation issues. In the assessment of developmental toxicity data, statistical

Criteria for developmental toxicity * D. R. MATTISON ET AL. considerations require special attention. One impor- tant area is the power of the study, which is the prob- ability that the study will demonstrate a true effect. It is dependent on the sample size as well as the back- ground incidence and variability of the endpoint(s) examined. The apparent lack of an effect on a specific endpoint may be due to a true lack of activity or the inability of the study to identify an effect because of small sample size. Conversely, some statistically sig- nificant effects may arise by chance, especially if a large number of endpoints are analyzed. The Expert Committee encourages appropriate statistical evalua- tion of studies. Even with appropriate statistical tests, some statistically significant differences may arise by chance alone; the use of appropriate historical control data may prevent a false assumption of biological rel- evance in such cases. 3. Role of maternal toxicity in developmental toxicity. One consideration in the identification of an agent as a developmental toxicant is whether it pro- duces adverse effect(s) on the conceptus at exposure levels that do not also induce severe maternal toxicity (e.g., substantial reduction in maternal weight gain, persistent emesis, hypo- or hyperactivity, or convul- sions). The main reason for conducting developmen- tal toxicity studies is to ascertain whether an agent causes specific or unique toxic effects on the concep- tus. If these studies are conducted under conditions of extreme maternal toxicity, then identification of expo- sure uniquely toxic to the conceptus or pregnant ani- mal is difficult. As previously discussed, one criterion for iden- tifying a developmental toxin is determination of the relative toxicity of the substance to the adult mother and the developing conceptus. In humans there ap- pear to be substances that are toxic to conceptuses in the absence of apparent maternally toxic effects (e.g., thalidomide, diethylstilbesterol, ionizing radiation) and substances that are toxic to the conceptus at con- centrations that are used therapeutically or that result in maternal physiologic changes (e.g., tobacco, steroi- dal hormones, alcohol, methylmercury, 13-cis-retinoic acid, phenytoin, and valproic acid). Based on its re- view of the literature, the Expert Committee recom- mends that greater concern be given to developmental effects identified to be dose-dependent and present at doses that do not also result in maternal toxicity. V. SUMMARY Because of the automatic restrictions and warn- ing requirements imposed on substances identified by the state as "known to cause developmental toxicity," the Expert Committee recommends the use of criteria that emphasize human relevancy, biological plausibil- 11 ity, and evidence in support of a selective, adverse developmental effect at non-maternally-toxic doses. In many instances, data for substances of public concern will be insufficient at present to meet these criteria. The fact that a substance is not listed as "known to cause developmental toxicity" does not cre- ate a presumption that the substance is safe. The Expert Committee, therefore, urges that these sub- stances be recommended for further testing and that high priority be given to conducting the necessary tests. The Expert Committee reiterates its concern that substances listed by the SAP be identified according to the toxic endpoints (cancer, male reproductive tox- icity, female reproductive toxicity, and/or develop- mental toxicity) that led to listing. Further, the Expert Committee recommends that the state Health and Welfare Agency institute education programs empha- sizing appropriate courses of action for citizens in- formed of exposures to substances known to the state to cause cancer, birth defects, or reproductive toxicity. REFERENCES 1. California Health and Welfare Agency 1987. Chemicals known to cause cancer or reproductive toxicity. March 1, 1987. Sac- ramento: State of California Health and Welfare Agency. 2. Safe Drinking Water and Toxic Enforcement Act of 1986. Scientific Advisory Panel, Reproductive Toxicity Subpanel, Draft for comment, December 11, 1987, State of California Health and Welfare Agency. 3. EPA (Environmental Protection Agency). Guidelines for the Health Assessment of Suspect Developmental Toxicants. Fed- eral Register. 1986;51(185):34028-40. 4. NAS (National Academy of Scipnces). Risk assessment in the federal government: managing the process. Washington: Na- tional Academy Press; 1983. 5. Nebert DW, Atlas SA. The Ah locus: aromatic hydrocarbon responsiveness of mice and men. Hum Gen. 1978;1 (Suppl.): 149-60. 6. Omenn GS, Gelboin HV, eds. Genetic variability in responses to chemical exposure. (Banbury Rept. 16.) Cold Spring Harbor, NY: Cold Spring Harbor Laboratory; 1984. 7. Brown NA, Freeman SJ. Alternative tests for teratogenicity. Alternatives Lab Anim. 1984;12:7-23. 8. Fabro S, Schull G, Brown NA. The relative teratogenic index and teratogenic potency: proposed components of developmen- tal toxicity risk assessment. Teratogenesis Carcinog Mutagen. 1982;2:61-76. 9. Johnson EM. Screening for teratogenic hazards: nature of the problem. Annu Rev Pharmacol Toxicol. 1981;21:417-29. 10. Brent RL. Editor's note. Teratology. 1978;17:183. 11. Shepard T. Human teratogenicity. Adv Pediatr. 1986;33:225-68. 12. Stein Z, Kline J, Kharrazi M. What is a teratogen? In: Kalter H, ed. Epidemiologic criteria issues and reviews in teratology (vol. 2). New York:Plenum Press; 1984. 13. Wilson JG. Embryotoxicity of drugs in man. In: Wilson JG, Fraser FC eds. Handbook of teratology. New York:Plenum Press; 1977. 14. Spranger J, Benirschke K, Hall JG, et al. Errors of morpho- genesis: Concepts and terms. J Pediatr. 1982;100:160. 15. Marois M, ed. The significance of minor malformations: syn- drome identification. International Conference on Prevention of Physical, Mental, and Congenital Defects sponsored by the 9

r ~ 12 Reproductive Toxicology Volume 3, Number 1, 1989 Institut de la Vie. Strasbourg, FranCe (October 1982). New York:Alan R. Liss; Part C, 1982;39-70. 16. Miller RW. The discovery of human teratogens, carcinogens, and mutagens: lessons for the future. In: Hollaender A, deSerres FJ, eds. Chemical mutagens: principles and methods for their detection. New York: Plenum Publishing Co.; 1978:101-26. 17. Bloom AD, ed. Guidelines for studies of human populations exposed to mutagenic and reproductive hazards. White Plains, NY: March of Dimes/Birth Defects Foundation; 1981. 18. Hennekens CH, Buring JE. Epidemiology in medicine. Boston: Little, Brown & Company; 1987. 19. Wang GM, Schwetz BA. An evaluation system of ranking chemicals with teratogenic potential. Teratogenesis Carcinog Mutagen. 1987;7:133-9. 20. FASEB. Predicting neurotoxicity and behavioral dysfunction from preclinical toxicologic data. Bethesda, MD: Life Sciences Research Office, Federation of American Societies for Exper- imental Biology; May 1986.