Toxicology and Industrial Health

This paper describes EPA's recent efforts to ensure consistency and technical competence of the agency's risk assessments. These include five risk assessment guidelines proposed in November 1984 and January 1985 and establishment of a Risk Assessment Forum. The technical provisions of the five guidelines have been outlined.

This paper describes two general methods for estimating ADIs that circumvent some of the limitations inherent in current approaches. The first method is based on a graphic presentation of toxicity data and is also shown to be useful for estimating acceptable intakes for durations of toxicant exposure other than the entire lifetime. The second method uses dose-response or dose-effect data to calculate lower CLs on the dose rate associated with specified response or effect levels. These approaches should lead to firmer, better established ADIs through increased use of the entire spectrum of toxicity data.

Predictions of human health risk for single chemicals are often based on animal studies and hence require some sort of adjustment for species differences in toxic susceptibility. In the past, either the animal dose has been divided by an uncertainty factor or the dose has been transformed by a mathematical model into a human equivalent dose. A generalization of the allometric model previously used for carcinogens, the so-called "surface area model, " is investigated here for use with graded severity response data for noncarcinogenic systemic toxicity. Statistical methods for estimating one of the model's parameters, the power of body weight, are proposed and tested on simulated and actual toxicity data. Early results indicate reasonable accuracy if data are available for a large number of dose groups.

Epidemiologic data with quantitative exposure measures is infrequently available for specific environmental agents. This lack of exposure measures creates confusion in interpreting epidemiologic data and therefore has impeded its efficient use in health risk analysis. This paper discusses screening and evaluating epidemiologic studies for use in assessing health risk. It also describes the larger role of epidemiology in reducing uncertainties in risk analysis. The approach recognizes that the various designs used to increase statistical power and to control for covariables have different functions in contemporary risk assessment as practiced by regulatory agencies. Each of these study designs is categorized for its role in risk analysis as useful for hazard identification or for dose-response assessment. Studies presenting geographic correlations are construed to be not directly useful in health risk assessment. The numerical level of the exposure data is a deciding factor in using valid epidemiologic studies. However, data measured on an ordinal scale can be used in qualitative assessments and can demonstrate the strength of the relationship. The application of this procedure is illustrated using epidemiologic studies on the carcinogenicity of chemicals contaminated with dioxins.

This paper briefly discusses criteria for evaluating epidemiologic studies for risk assessment purposes, using asbestos as an example. Asbestos is one of the few carcinogens for which substantial data exist on exposures to humans. However, there are major difficulties in using these data for conducting risk assessments. In particular, exposure data are often incomplete, and risk assessments usually involve extrapolating from the higher exposures of the occupational environments to the lower levels typically encountered in the nonoccupational environment. The term "asbestos" refers to the fibrous form of several minerals, and levels of exposures to these fibers are not easily assessed. Criteria for evaluating epidemiologic studies used in an Ontario Royal Commission report on asbestos are discussed. The importance of considering the statistical power of studies to detect an excess risk is examined using as examples major cohort studies of asbestos-exposed workers, as summarized in a report by the U. S. National Research Council.

The process of assessing the risk associated with human exposure to environmental chemicals inevitably relies on a number of assumptions, estimates and rationalizations. One of the more challenging aspects of risk assessment involves the need to extrapolate beyond the range of conditions used in experimental animal studies to predict anticipated human risks. The most obvious extrapolation required is that from the tested animal species to humans; but others are also generally required, including extrapolating from high dose to low dose, from one route of exposure to another and from one exposure timeframe to another. Several avenues are available for attempting these extrapolations, ranging from the assumption of strict correspondence of dose to the use of statistical correlations. One promising alternative for conducting more scientifically sound extrapolations is that of using physiologically based pharmacokinetic models that contain sufficient biological detail to allow pharmacokinetic behavior to be predicted for widely different exposure scenarios. In recent years, successful physiological models have been developed for a variety of volatile and nonvolatile chemicals, and their ability to perform the extrapolations needed in risk assessment has been demonstrated. Techniques for determining the necessary biochemical parameters are readily available, and the computational requirements are now within the scope of even a personal computer. In addition to providing a sound framework for extrapolation, the predictive power of a physiologically based pharmacokinetic model makes it a useful tool for more reliable dose selection before beginning large-scale studies, as well as for the retrospective analysis of experimental results.

Normal human subjects vary widely in their capacity to eliminate many drugs and environmental chemicals. These variations range in magnitude from fourfold to fortyfold depending on the drug and the population studied. Pharmacogenetics deals with only one of many host factors responsible for these large pharmacokinetic differences. Age, sex, diet and exposure to other drugs and chemicals, including oral contraceptives, ethanol and cigarette smoking, can alter the genetically determined rate at which a particular subject eliminates drugs and environmental chemicals. These elimination rates, therefore, are dynamic and change even in the same subject with time and condition. Regulatory legislation has only recently begun to recognize this very broad spectrum of human susceptibility and the existence of multiple special subgroups of particularly sensitive subjects. In setting standards for environmental chemicals, EPA and NIOSH have attempted to protect the most sensitive humans and should be encouraged to continue this policy. For some drugs and environmental chemicals, the commonly used safety factor of 100 may be too low; for these chemicals large, interindividual pharmacokinetic variations produced by pharmacogenetic and other host factors may make a safety factor of 400 or 500 more adequate.

Results of acute toxicity studies for a variety of chemicals have indicated that, in most cases, although the inhalation route was more effective than the IG route, wide variations in toxicity occurred between these two routes. The major factors that may result in variations in toxicity between routes include: (1) differences in absorption efficiency; (2) differences in systemic effects; (3) occurrence of critical toxicological effects at the portal of entry; (4) first-pass effects resulting in inactivation or activation of the chemical agent before it reaches the target organ; and (5) variations in temporal patterns of target organ concentrations.

Extrapolation to determine safe exposure levels during chronic exposure becomes less reliable, not only as information relating to these factors decreases, but also as the quality or length of exposure decreases in the available toxicologic studies. VDC is an example of one of a few chemicals for which both chronic inhalation and oral toxicity data are available, along with detailed pharmacokinetic information. On the basis of the pharmacokinetic data, differences in toxicity between the two routes did not appear to be very likely for this chemical. This conjecture was supported by the results of chronic toxicity studies. Finally, assuming sufficient data and pharmacokinetic parameters are available, this paper presents a useful and practical approach to route extrapolation.

During the past 3 years, EPA's ECA O/ Cincinnati has developed a method to rank chemicals based on chronic toxicity data. This ranking system reflects two primary attributes of every chemical: the minimum effective dose and the type of effect elicited at that dose. The purpose for developing this chronic toxicity ranking system was to provide the EPA with the technical background required to adjust the RQs of hazardous substances designated in Section 101(14) of CERCLA or "Superfund." This approach may have applications to other areas of interest to the EPA and other regulatory agencies where ranking of chemicals based on chronic toxicity is desired.

A comparative potency method for cancer risk assessment has been developed based on a constant relative potency hypothesis. This method was developed and tested using data from a battery of short-term mutagenesis bioassays, animal tumorigenicity data and human lung cancer risk estimations. This data base was developed for a series of complex mixtures including emissions from coke ovens, roofing tar pots, cigarette smoke and automotive engines. The series of automobiles used in this study included both diesel- and gasoline-powered vehicles. The assumptions inherent in this method are discussed, together with the methods and data base used to test these assumptions.

Clarity and scientific validity are two criteria for assessing the quality of communications between scientists and risk managers. Regulating permissible exposures on the basis of very-low-dose risk extrapolation uses scientific information that may not meet either criterion. With regard to clarity, it is difficult for an individual to conceive of the meaning of risks on the order of 1 in 1,000,000 lifetime excess. With regard to scientific validity, the uncertainties of extrapolating risks at very low doses are evident in the wide variation in results produced depending on the statistical and pharmacokinetic assumptions made. An alternative approach is to fix benchmarks from which safety factors are chosen (Gaylor, 1983). For epidemiological data, we propose a benchmark of that exposure which would cause 1% cancer excess with 10 yr of exposurefollowed by 30 yr of further followup. These values have been chosen at this stage of development since they are close to observable values in many cancer epidemiology studies. In the first stage, excess cancer risk versus duration of exposure is plotted and the excess risk is estimated for a exposure duration of 10 yr. The next step involves a short linear extrapolation from the exposure levels linked with this excess risk to the exposure levels that would cause a 1% excess risk over a duration of 10 yr. We have established preliminary benchmarks from published data for benzene and radon daughter exposure. Permissible exposure levels might then be set by deciding on safety factors. Although low-dose extrapolation will aid in selecting a safety factor for carcinogenic exposures, it should be acknowledged that the actual risk at the permissible level is not known. The risk might be zero, it might be between zero and a linear extrapolation value, or it might possibly be a little higher than the value from a linear projection (Gaylor and Kodell, 1980; Gaylor, 1983).

Many chemicals are of concern to human health, but only a few have epidemiologically derived risk estimates. About 45,000 chemicals are listed in RTECS, most of which have had some testing in subhuman models. RTECS entries range from cellular effects through organoleptic damage to lethality, with many pathological endpoints listed, including mutagenic changes, irritation, teratogenesis, cancer, mortality, etc. However, it is difficult to extend any biological test results to human risk assessments. If the results are extended, the degree of validity is highly uncertain.

This paper describes a logical basis for using the entire complex spectrum of test results to evaluate the overall toxicological potency of a chemical to be assayed (i.e., an interviewing chemical) and describes how to derive tentative, permissible concentrations in air and water for any particular chemical for which no regulatory guidance exists. This approach has been tested for 16 reference chemicals discussed in NIOSH Criteria Documents, EPA-CAG reports, etc. The evaluations are uncomplicated, but occasionally it is difficult to match RTECS entries for two different chemicals. Difficult comparisons may require some familiarity with experimental design and the toxicological literature.

One important product of this novel approach is that a distribution or array of potency values is obtained for any chemical evaluated. This distribution reflects many uncertainties stemming from low statistical power, experimental design, pharmacological processes, interspecies variability, dose rate, biological effect monitored, route of treatment, etc. The array of relative values for a particular chemical reflects many different biological and physical conditions. The distribution of the array helps to index a composite toxicological profile for many different biological effects resulting from numerous treatment protocols.

To minimize the effect of extreme sensitivity of certain (perhaps novel) biological test models, possible errors in the RTECS database, and possible human pharmacological insensitivity to a particular chemical and/or a particular route of administration, we consider the interquartile range (i.e., the central 50%) of the array of relative potency values between two chemicals being compared as a practical measure of uncertainty. Thus, the range in response derived from variability in relative potency should be useful in addressing the range of response in man as estimated from extrapolations of test data.

EPA is faced with the need to perform health risk assessments of environmental materials containing mixtures of chlorinated dioxins and -dibenzofurans. Preferably, such an assessment must be based on the direct evaluation of chronic health effects of the mixture or on a knowledge of the toxic effects of each of the components. In the absence of such data, the authors have developed two pragmatic approaches that can reasonably be applied to mixtures such as flyash or contaminated soil. The details of these approaches, viewed as interim procedures, are compared and contrasted with those used by other regulatory and public health authorities.

The use of quantitative structure-activity relationships (QSAR) is considered with respect to estimating the carcinogenic risk of untested chemicals. SAR derived from a retrospective classification of a series of aromatic amines were used to study the estimation of carcinogenic risk by analogy. Using pattern recognition methods, a series of molecular descriptors were developed for a data set of aromatic amines that supported a linear discriminant function capable of separating compounds testing positively for carcinogenicity from those testing negatively. Linear discriminant analysis correctly categorized the compounds as positive or negative in 94.9% of the cases. For each aromatic amine within the subset of positive compounds, the most appropriate analogue was identified using physicochemical, topological, geometric and electronic molecular descriptors as variables. An upper-limit unit risk estimate was calculated for each compound that was a positive carcinogen within the data set using the linearized multistage model. The actual risk and the risk estimated by analogy to a congener were compared for each compound within the positive subset. The results support estimating both qualitative and quantitative carcinogenic risk by analogy for this particular data set.

Acceptable daily intake (ADI) values are routinely developed for threshold toxicants from NOAELs determined from human or animal chronic or subchronic data. These NOAELs are then divided by appropriate uncertainty factors ranging from 10 to 1000 depending on the quality of the data. However, for the vast majority of chemicals used industrially, adequate toxicity data needed to use this process are not available. Thus, a procedure to estimate a chronic toxicity endpoint from acute toxicity data, such as an oral rat LD₅₀, becomes necessary. An acute-to-chronic application factor of 0.0001 was developed, which when multiplied by an oral LD ₅₀ for an individual chemical, yields a surrogate chronic NOAEL. This figure can then be used to estimate an acceptable daily exposure for humans. The process used to estimate this application factor is detailed.

This paper examines the issues involved in health risk assessments at hazardous waste sites and resource recovery facilities in Region II and emphasizes the programmatic aspects of the problem and the integration of risk assessment into the permitting process for these facilities. Suggestions for the R&D agenda and policy needs to support a risk assessment strategy in this area are also discussed.

This report provides an example of how a single source of contamination could potentially contribute to all routes of exposure. A modeling approach was used to estimate multiple exposure routes in an attempt to assess the health significance of gasoline-contaminated drinking water supplies. This model consisted of a two-compartment, indoor air quality equation that calculates the contribution made by ambient and indoor air contaminated by a pollutant volatilized from drinking water to that pollutant's inhalation burden. In addition, the model uses the traditional equations for assessing a pollutant's oral and dermal burdens. Benzene, toluene and xylene were used as surrogates for gasoline contamination to determine the contribution of contaminated water to adult and child body burdens from indoor air, oral (drinking water and food) and dermal exposure routes. The contribution thus calculated for each chemical was compared to the EPA's Office of Drinking Water Health Advisories.

In terms of acute exposure, the use of chemically contaminated water for showering purposes may generate vapor in the confined area of the bathroom at levels sufficient to cause or contribute to mucous tissue irritation, as commonly reported in affected homes. High temperatures and humidity may also contribute to these effects, especially in the bathroom. In terms of chronic exposure, the use of chemically contaminated water at EPA-recommended guideline amounts in an affected home may result in inhalation, oral and dermal exposures leading to cumulative doses exceeding adult and child total daily body burdens based on EPA's Health Advisories. Thus, this model indicates that the traditional standard/guidelines derivation processes should be reevaluated to consider the pollutant contribution from multiple routes of exposure.

The New Jersey Departments of Health and Environmental Protection conducted a study in which concentrations of several pollutants including benzene in the breathing zone were measured during a 15-minute shower in homes with contaminated water. The findings suggest that the air quality model used in the present study may satisfactorily predict the airborne concentrations of pollutants in, at least, the bathroom after showering with contaminated water (Pearson rank correlation coefficient of 0.773 with p = 0.0012 for n =14).

The findings of the present study support the use of an adjustment factor for all exposure durations to account for exposures to other sources of the contaminant, i.e., urban, occupational, and food. A value of 20% seems appropriate based on the study's findings. However, it is more likely that this value may be too high considering that occupational and food exposures for most of the contaminants evaluated in the study were not accounted for.

The generated values in this study are theoretical and, therefore, should be cautiously interpreted. Limitations of the model include inappropriate use of steady-state dermal absorption rates for the pollutants considered in this study, poor air stripping data, and background indoor air quality data, among others. However, given the limitations of the model, the inherent conservative approach employed in calculating body doses and drinking water guidelines, and the scant reports of ill health effects in households with gasoline-contaminated drinking water, the findings presented in this report should not be discounted and should suggest research efforts to confirm or refute the findings presented in this report. Primary research efforts should concentrate on evaluating dermal uptake of pollutants under initial rate conditions in varying temperatures of soapy water.

The Maine Departments of Human Services and Environmental Protection are gathering case history information on several selected homes with contaminated drinking water that have not yet installed charcoal filters. The information gathered will include the results of air monitoring in the bathroom with the shower flowing, an assessment of the pattern of contaminated water use and health questionnaire, among others. The brief health questionnaire will concentrate primarily on acute effects to assess irritant potential of gasoline vapors in bathrooms. Similar information will be gathered from control homes for comparison.

This paper reviews the problem of performing risk assessments using data on fetal toxic effects. It briefly discusses the usual dose-response models and their inappropriateness for application to such data. The paper then considers tests for determining whether the fetal toxic effect is increased over that of the control group. Assuming an increase has been shown, the use of a fetal toxicity, dose-response model for risk assessment is discussed. The paper then applies these methods to data from an experiment using female mice mated with irradiated males. Finally, the paper discusses the need for further statistical research in this important area.

Significant numbers of chemicals have been shown to be carcinogenic in mouse liver although they do not exhibit carcinogenic activity in other organs or tissues of mice or rats. This review focuses on the reasons for the unique susceptibility of the mouse liver to these carcinogens and the extent to which the carcinogenic activity of a chemical in mouse liver can be used to predict carcinogenicity in humans.

Many of these mouse liver carcinogens lack genotoxic activity and, as such, have been proposed to be tumor promoters. Two mechanisms that may explain the action of nongenotoxic carcinogens in mouse liver are reviewed. These are: (1) direct action on precursor cancer cells, either to accelerate their growth or to prevent their death and (2) the selective growth advantage, resulting from regenerative hyperplasia of precursor cancer cells in response to the necrosis of normal cells produced by hepatotoxins.

Estimating human health risks on the basis of mouse liver tumor data is believed to differ for nongenotoxic and genotoxic carcinogens in two fundamental ways. The first involves intraspecies extrapolation and the second involves low-dose extrapolation. In conclusion, although mouse liver tumor data are seen to be of value in estimating human health hazard, it is important to distinguish between genotoxic and nongenotoxic mechanisms in applying such data. Further study of the biochemical and molecular mechanisms of chemical carcinogens is necessary to determine the relationship between their activity in mouse liver and their activity in humans.