WORKSHOP ON RISK AND DECISION MAKING EPA Office of Solid Waste Hyannls Regency Inn Hyannis, Massachusetts December 8 and 9, 1986 ------- The materials presented here have been reviewed by personnel from the United States Environmental Protection Agency. They do not, how- ever, necessarily reflect United States Environmental Protection Agency policy. The materials were prepared primarily by: TEMPLE, BARKER & SLOANE, INC. Lexington, Massachusetts ENVIRON CORPORATION Washington, D.C. ------- WORKSHOP ON RISK AND DECISION MAKING I &EPA Office of Solid Waste Hyannis Regency inn - Hyannis, Massachusetts December 8 and 9, 1986 ------- MEMORANDUM TO: Participants in the Office of Solid Waste’s Case Study on Risk and Decision Making RE: Purpose of the Case Study and Your Role Risk assessment is becoming an increasingly important compo- nent of regulatory decision making within EPA. This is true for a number of EPA offices, including the Office of Solid Waste. Even so, the components and results of a risk assessment are often not well understood, and the specific ways in which risk assessment can be used in a particular situation are not well defined. The result is confusion and ultimately frustration among those who must carry out or are affected by the regula- tions. Risk assessments, however, need not be overwhelming and can provide guidance in t hinking about an environmental problem. The purpose of this workshop is to help individuals in the EPA Regions understand the basis of risk assessment and to develop a common base of knowledge and terminology. The workshop is not intended, however, as a step—by—step “how to” on risk assessment. Rather, we are trying to explain the concepts of risk assessment in the context of a hypothetical case study. With any risk assessment, there are many uncertainties and issues, and often no clear “right” answers. As you proceed through the case, you will be asked to consider those issues. Some of the issues you normally would not confront in your day—to-day activities, and some of the terms may be new. But, regardless of your back- ground, we hope you will address those issues and develop your own conclusions. By helping you understand the issues, we hope to make you more informed consumers of Information on risks and increase your understanding of both the potential and the limita- tions of risk assessments. In addition, we hope to encourage discussion of the unique circumstances involved in developing risk assessments for the site—specific, localized problems that confront individuals dealing with Resource Conservation and Recovery Act (RCRA) issues. The workshop case study will focus on a hypothetical site— specific problem that EPA regional staff might encounter in their day—to-day work under RCRA. You will have to decide what actions should be taken at a site regulated under RCRA. ------- 2 BACKGROUND ON THE CASE The Company Electrobotics was formed 15 years ago as a privately held company. Its two owners have lived in the community for many years. Electrobotics manufactures electronic components for the computer industry. In the past few years, the bulk of the com- pany’s sales has been in parts for personal computers, particu- larly the popular Bananachrome personal computer. Electrobotics has had a rocky history and has been close to bankruptcy several times. However, with the recent growth in demand for personal computers, its performance has improved. Last year Electrobotics had revenues of $7.5 million, compared with $1.5 million five years ago. Even so, the company was only marginally profitable, with after—tax profits of less than $50,000. Electrobotics employs about 150 people and has a reputation for treating its employees well. It has never laid off an employee, even during difficult periods. Location Electrobotics is situated in a sparsely populated region. A residential development was recently built near the facility’s western boundary, and approximately 80_ op1e live in2O_homes there. In addition, the town of j t 1a—lies a little more than 1 mile east of the facility. The area between the facility and — — - - the town has few homes, none closer than three quarters of a mile to the plant’s eastern boundary. There has been some discussion about building housing on the property, but no definite plans exist at this time. Utopia relies o gx.Qun&. ater for its public water_system and has 40 wells in a nearby well field. The welF — — ---. field supplies including those in the development west of Electrobotics. The ground water from the Electrobotics site flows toward the public weTrTiird. mrtr i the Electrobotics site would affect only_three—of—the-_p..ublic_wells. (Figure 1 provides a schematic of the Electrobotics site aiW€he surrounding area.) ------- 3 Figure 1 Site Plan Elactrobotics Property Boundary — — — ROAD -J 1000 ft.— ’ Residential Area U- I I I I I I L .1 0 7.000 ft. - 0’ Direction of Ground-Water Flow N (Not to Scale) Nature of the Problem About five years’ ago, E].ectrobotics changed its production process as it introduced a new product. The process required the use of two industrial solvents-—dinitrochickenwire (DNC) and SOLVALL. Wastewater containing the spent solvents is stored in a flow equalization lagoon that is part of the Electrobotics waste— water treatment plant (see Figure 1). After treatment, the wastewater is discharged to the local publicly owned treatment works. Electrobotics’ treatment plant uses an activated sludge process. As a result, no concentrations of DNC or SOLVALL.we.r found during initial ment facility,_ and no DNC or SOLVALL concentrations were foundii tE ai T i’ Tde the treatment plant. EPA has classified the two solvents as hazardous wastes. Information on the carcinogenic risks from SOLVALL was published in 1981. Studies on the carcinogenic risks from DNC, however, were just recently presented. These will be discussed in more detail as you assess the risks at the site. \ Threatened Water Supply Wells ------- 4 STATUTORY AND REGULATORY ISSUES The hazardous waste regulations developed under RCRA require that owners and operators of hazardous waste facilities, like Electrobotics’ storage lagoon, utilize design features and con- trol measures to prevent the leaking of hazardous waste into ground water. Furthermore, all regulated units (a regulated unit, such as the storage lagoon, is a facility that received hazardous waste after July 26, 1982) are subject to monitoring and corrective action requirements if a ground-water problem is identified. The ground-water protection standards require the Regional Administrator to establish in the facility permit, for each hazardous constituent entering the ground water from a regu- lated unit, a concentration limit beyond which degradation of ground-water quality is not allowed. The concentration limits determine when corrective action is required. Three possible concentration limits can be used to establish the ground-water protection standards: • Background levels of the hazardous constituents • Maximum concentration limits (MCLs) established by regulation • Alternate concentration limits (ACL5) Background levels and MCLs are established in the facility permit unless the facility owner or operator applies for an PCL. The background levels of DNC and SOLVALL are zero, and no MCLs exist for the two solvents. In October 1985, the owners of Electrobotics submitted to EPA, as required under the RCRA regulations, a Part B application for an RCRA permit. This application provides the Regional Administrator with information necessary to evaluate the safety of the site. It generally includes: • A description of the facility • Chemical and physical analyses of the hazardous waste to be handled • Information from a hydrogeological investigation of the site - • The potential for the public to be exposed to hazardous wastes or hazardous constituents through releases related to the unit, including identification of the ------- 5 potential pathways of human exposure and the magnitude and nature of the human exposure resulting from such releases In addition, Electrobotics’ owners have been conducting ground—water monitoring as required under the regulations. The results presented in the Part B application and from other sources indicate that both DNC and SOLVALL were found in th ground. wfer.—be-l-ow---t-he_ lagoon and in the air . The initial monitoring wellswere locate ETh ii dThEčëd’ge of the lagoon, referred to as the point of compj jan e. ” Additional monitoring wells have been irii alled to characterize the plume, and results indicate that it has moved to the eastern edge of the facility boundary. YOUR ROLE You, as the EPA representative responsible for the Electro— botics site, will evaluate the available information on the risks posed by DNC and SOLVAt 1 L and decide whether to issue a permit and what, if any, corrective actions should be taken at the facility. Part I During Part I, you wish to determine whether and to what extent the DNC and SOLSIALL found at the Electrobotics plant endanger the public health. Thus, you have decided to conduct a risk assessment . Your staff has prepared a document summarizing much of the information you will need for assessing the risks. This document, attached as Part I, reviews the potential risks from DNC and SOLVALL. The first section, which contains toxicological data on the two solvents, constitutes the hazard evaluation . The second section of Part I contains a summary of data on the exposure of various population groups to DNC and SOLVALL. Several issues arise concerning the interpretation and use of this information, and it will be necessary for you to formulate appropriate conclusions. ------- 6 The relationship between exposure to DNC and SOLVALL and the risk of adverse health effects is determined by a dose—response evaluation . In the third section of Part I, you must review possible approaches to defining this relationship for the two solvents. There may be several plausible options for describing this relationship in the region of human exposure. Normally, you would not examine the hazard evaluation data and the dose- response information; these results would be provided by EPA headquarters. However, in this instance you have asked your staff to provide the material for your review because you feel it is important to your understanding of the problem. In the final section of Part I, you must present your con- clusions regarding the human health risks posed by DNC and SOLVALL from the Electrobotics facility, and the uncertainties in your knowledge. You must characterize the risks . Part I will be completed during the first day of the workshop. Part II Once you have assessed the risks from DNC and SOLVALL, you must decide what actions should be required by Electrobotics’ owners. You must manage the risks . Part II of the case is currently being prepared by your staff. It will be given to you at the end of the first day of the workshop, when you have completed your assessment of the risks, and will be discussed the next morning. The new document will contain information on the options for cleaning up the storage lagoon at the Electrobotics site. In particular, you will have an approach proposed by the owners of Electrobotics and two alternatives identified by your staff. For each of the four options, you will be presented with information describing the approach, predicting the effect of the option on the risks identified in Part I, and estimating the costs. At each of the steps in Part I and Part II, issues and data will be presented and alternative conclusions listed. After discussion, you may select the conclusion that seems most appro- priate; if none seems appropriate you should offer your own. Your review and evaluation will take place within a working group of 10 to 12 people. After you discuss the issues with the other members, your group should attempt to reach a consensus on what regulatory action should be taken. If you cannot reach a ------- 7 consensus, present the alternative views. You should note, however, that you must ultimately decide what actions will be required of Electrobotics’ owners. You will be meeting with the Regional Administrator at the end of your session and must present your recommendation to him at that time. ------- Part I ASSESSING THE RISKS FROM DNC AND SOLVALL ------- CONTENTS I. BACKGROUND ON THE CHEMICALS DETECTED AT THE ELECTROBOTICS SITE II. HAZARD EVALUATION III. HUMAN EXPOSURE EVALUATION IV. DOSE-RESPONSE EVALUATION V. RISK CHARACTERIZATION GLOSSARY APPENDIXES A. Ground-Water Modeling Calculations and Associated Assumptions B. Human Dose Calculations and Associated Assumptions ------- I. BACKGROUND ON THE CHEMICALS DETECTED AT THE ELECTROBOTICS SITE Dinitrochickenwire (DNC ) • Solvent used to degrease fabricated metal parts • Impurities: commercial product contains trace amounts of trinitrochickenwire • Physical state: liquid, slight to moderate volatility • Stability: degrades very slowly in aqueous environ- ments • Solubility: moderately soluble in water SOLVALL (dichioropentagon ) • Solvent used in the production of microchips • Physical state: liquid, moderate to high volatility • Stability: degrades very slowly in aqueous environ- men t s • Solubility: highly soluble in water ------- II. HAZARD EVALUATION Normally, you would not review information on the hazards of specific chemicals; the results would probably be provided to you by EPA toxicologists. But by working through the material in this section, you will develop an understand- ing of the nature and quality of information used by the toxicologists. You will not become toxicologists, but we hope you will become more informed users of the information they provide. ------- 11—2 SOME PRINCIPLES FOR HAZARD EVALUATION • The purpose of hazard evaluation is to identify the types of adverse health effects that may be associated with exposure to DNC and SOLVALL, and to characterize the quality and strength of evidence supporting this identification. • The specific hazard of concern in this review is cancer. • Epidemiological studies in exposed human populations are generally considered the best source of inforatatiori for hazard identification. Unfortunately, they are not: available for most substances. Moreover, establishing firm causal links between exposure and chronic human disease (such as cancer) is very difficult. • Studies with experimental animals also provide useful information for hazard identification. Such studies can be controlled, and thus can more easily establish causality. Results from such studies suffer from the obvious limitation that experimental animals are not the species of ultimate interest. • With one possible exception (arsenic), all known human carcinogens are also carcinogenic in one or more species of experimental animals. Most animal carcino- gens have not been established as human carcinogens. In most cases, the lack of adequate epidemiological data means that no decision about human carcinogenicity can be made. • Biological data support the proposition that responses - in experimental animals should be mimicked in humans. For some agents, however, species differences in response can be substantial. • The specific sites of tumor formation in humans may differ from those observed in experimental animals. • Data obtained by administering a substance by the same route of exposure experienced by humans are considered more predictive than data obtained by a different route. If tumors form at internal body sites, however, the route of exposure may not be important. ------- 11—3 • In general, a varied response in experimental animals—— tumor formation in several species and both sexes, at different exposure levels with increasing response at increasing exposure, and at multiple body sites—— provides more convincing evidence of potential human carcinogenicity than does a response limited to a single species or sex, or to body sites at which tumors commonly occur in untreated animals (e.g., liver tumors in untreated female mice). BACKGROUND ON DNC TOXICITY The toxic properties of DNC were first investigated in the 1940s and 1950s. In most of these tests, small groups of experi- mental animals were exposed to very high amounts of DNC to iden- tify the exposure conditions that would cause death. Animals received either a single exposure, or exposures covering only a fraction of their lifetime. During the 1950s and l960s, more extensive animal toxicity tests were conducted, although none involved DNC exposures last- ing more than about one—sixth of a lifetime. These tests revealed the range of doses that produced toxicity (the principal site of toxic action was the liver) and the exposure level below which no form of toxicity was identified. Information available in 1970 showed that the most highly exposed humans received a daily DNC intake several hundred times lower than the “no—observed toxic effect” intake identified in the animal tests. Because the liver toxicity produced by DNC was of a type likely to occur only ‘after a minimum threshold exposure was exceeded, it was concluded that the most highly exposed humans were protected from DNC toxicity by a wide safety margin; that is, use of DNC presented no risk to the public health. At least until 1980, no data had been published on the effects of DNC on exposed humans. The Frankenstein Study In late 1985, an article titled “Chronic Toxicity of Dini— trochickenwire in Rats and Mice” appeared in a respected scien- tific journal (Frankenstein, V., 1. Environ. Tox.) . The Franken- stein paper presented data on the effects of lifetime exposure to DNC in two species of rodents. These data revealed a form of toxicity——carcinogenicity——that had not been seen before. The design of the Frankenstein experiment and the major findings are presented in Tables 1 and 2. ------- 11—4 Remarks on the Frankenstein Study 1. As far as can be determined from the published article, the Frankenstein study was carefully conducted, and there is no reason to doubt the accuracy of the reported data. 2. DNC increased the incidence of tumors (percentage or propor- tion of animals with tumors) in certain groups of animals. Not all animals in a group receiving DNC developed tumors. Tumor incidence is a measure of the risk (probability) of tumor development. Data in Table 2 can be interpreted as in the following example: the lifetime risk of stomach cancer in male rats exposed by gavage to the high dose of DNC daily, for their full lifetimes, is 0.40. 3. Rats developed spleen and liver tumors after both inhalation and gavage exposures. Lung tumors were produced only by inhalation, and stomach tumors only by gavage. Females of both species showed fewer tumors than males, and mice showed fewer tumors than rats. 4. The stomach tumors appeared at the point where DNC contacted the stomach when introduced by stomach tube. This point is in the rodent forestomach, an anatomical feature not present in humans. 5. Severe irritation was observed in the areas of the rodents’ stomachs that were exposed to high doses of DNC delivered by stomach tube. No signs of irritation were observed in low— dose control animals. Dr. Frankenstein believes that the stomach tumors arose as a result of the severe toxic insult caused by the direct stomach exposure. ------- 11—5 Table 1 DESIGN OF THE FRANKENSIEIN STUDY Number of Groups Animals Amount of Species end Receiving — ONC Received Duration of Route of Exposure DNC Male Fesale Each Day 1 Exposure (weeks) 2 Rat, inhalation Control 60 60 0 104 Lowdose 60 60 30 104 Highdass 60 60 60 104 Rat, gavsge 3 Control 60 60 0 104 Losdose 60 60 50 104 Highdoee 60 60 100 104 Mouse, gavege 3 Control 60 60 0 78 Londose 60 60 60 78 Highdoee 60 60 120 78 t Nilliqreme of OMC per kilogres of the animal’s body weight. The concentration of OMC in the air in the inhalation experiment has been converted to a unit of weight so it can be compared with the unite in the gavege study. 2 Approiuaate lifeepane of the animals under laboratory conditions. 3 Gevage is sdunistration of a substance by means of a stomach tube. Table 2 SIGNIFICANT FINDINGS FROM THE FRANKENSTEIN STUDY Following are the only groups in which a statistically eignificent excess of tumors was found. Nearly 40 possible sites of tumor formation were examined in each sex of both apecies. Percentage of Animals with Tumors (incidence rate) Tumors — — — — — Study Group Sex Found Control Low Dome High Dose Rat, inhalation Male Lung 3 25 Ret, inhalation Male Spleen 0 2 Rat, Inhalation Male Liver 3 7 l2 Rat, gavage Male Stomach 0 0 433 Rat, gavage Female Stomach 0 0 Rat, gavage Male Liver 3 7 15 a Rat, gavage Male Spleen 0 33 Mouse, gavmge Male Liver 5 30 a Mouse, gavags Male Stomach 0 0 9 A statistically significant excees of tumors relative to untreated control animals. This means it is unlikely that the difference in tumor incidence between the treated and control animals is due to chance. Because the only difference between the control end treated animals was the presence of DNC, it is likely that the excess tumor incidence is due to this compound. Tumors were found at other sitee in both control and treated animals, but no other tumors occurred in statistically significant excess. ------- 11—6 Issues to Be Considered 1. How do these data conform (or not conform) to the principles laid out on pages Il—i and 11—2——particu- larly the last principle? 2. In view of these principles, is there any reason to conclude that DNC is not carcinogenic in rats of both sexes (by inhalation and gavage exposures) and in male mice (by gavage)? 3. Should the stomach tumors be considered relevant to low—exposure risks to humans? 4. Should the data obtained by gavage treatment be considered relevant to human exposure? 5. Is there any reason to believe that humans would not be at risk of developing the various tumors, assuming exposure to DNC? 6. Is there any way to determine, from the data given, whether responses in humans are likely to be similar to those of rats or mice? Males or females? ------- 11—7 Epidemiological Data After the Frankenstein report was published, three DNC manu- facturers decided to submit reports to EPA on their investiga- tions into employee health. The information from these three reports is summarized below. Manufacturer A Manufacturer A had been producing DNC for 45 years, but none of the 161 employees in the mortality study had been exposed for more than 20 years; most were exposed for 10 to 15 years. Al- though employee exposure data were not extensive, they suggested that past exposures were relatively high, sometimes approaching the inhalation levels that produced an excess of tumors in animals. (Although the concentrations of air in the workplace approached those used in the animal experiment, the workers were exposed to these high levels for only a fraction of their life- time.) By January 1979, 35 of the 161 workers had died. No increase in cancers of any type was noted among these workers (3 cases observed, 3.8 expected in a population of the same size, sex, and age). Malignant neoplasms of the digestive system were elevated (2 cases observed, 0.7 expected), but this elevation was not statistically significant (i.e., it is not possible to say the observed difference was not due simply to chance). The woricers were also exposed to several other chemicals, at least two of which are known animal carcinogens. Manufacturers B and C Reports from Manufacturers B and C are similar to the report of Manufacturer A. No cases of malignant neoplasms of the stomach were reported by either manufacturer. Both manufacturers reported slight elevations in lung cancer, but neither elevation was statistically significant. No data on worker smoking habits were available. Manufacturer C reported 33 deaths among 290 employees. ------- 11—8 Issue to Be Considered Should the information from the DNC manufacturers alter earlier conclusions about the inferences to be drawn from the animal data? If so, how? If not, why not? ------- 11—9 EPA’S Risk Assessment Guidelines The Office of Health and Environmental Assessment (OHEA) within EPA’S Office of Research and Development has developed guidelines for carcinogen risk assessment. These guidelines discuss weighing the evidence that a substance is a carcinogen and classifying the chemical into one of five groups: Group A-—Human carcinogen Group B-—Probable human carcinogen Group C-—Possible human carcinogen Group D——Not classified as to human carcinogenicity Group E-—Evidence of noncarcinogenicity for humans OHEA developed an illustrative categorization of substances based on animal and human data, as shown in Table 3. Table 3 ILLUSTRATIVE CATEGORIZATION OF EVIDENCE BASED ON ANIMAL AND IIJMAN DATA Animal Evidence Hunan - No Evidence Sufficient Limited Inadequate No Data Evidence Sufficient A A A A A Limited Bi Bi Bi 81 81 Inadequate 82 C D D D Nodata 82 C 0 D E No evidence 82 C D D E ------- 11—10 Some Possible Conclusions About DNC Carcinogenicity 1. DNC is a human carcinogen (Group A). There is suff i— cient evidence from epidemiological studies to support a causal association between DNC exposure and cancer. 2. DNC is a probable human carcinogen (Group B2). There is sufficient animal evidence of carcinogenicity as demon- strated in the increased incidence of tumors at several sites in multiple species (rats and mice), in multiple experiments involving different routes of administration (inhalation, gavage), and at different dose levels. There is inadequate evidence of carcinogenicity from epidemiological studies. 3. DNC is a possible human carcinogen (Group C). There is limited animal evidence of carcinogenicity because the data obtained when DNC was administered by stomach tube may not be relevant to any route of human exposure. Thus, DNC resulted in increased tumors in only one species (rat) and in one experiment involving only the inhalation route of exposure. There is inadequate evidence of carcinogenicity from epidemiological studies. 4. DNC is not classifiable as to human carcinogenicity (Group D). Because of the extreme conditions under which tumors were produced in the animal experiments, there is no reason to believe that DNC is a possible human carcinogen. There is inadequate evidence of carcinogenicity from epidemiological studies. 5. Other (formulate your own conclusion). ------- ‘I—li BACKGROUND ON SOLVALL TOXICITY SOLVALL is a chemical that has received widespread accep- tance as an industrial solvent because it is relatively inexpen- sive and has excellent solvent characteristics. Since the 1930s, however, a number of reports have appeared in scientific litera- ture that document SOLVALL’s depressant action on the central nervous system. They include cases where humans were exposed accidentally for short periods to very high air concentrations of SOLSIALL, as well as the results of acute vapor exposures to SOLVALL in laboratory animals. Additional data on the toxicity of SOLVALL appeared in scientific literature during the 1960s. Several investigators reported the results of subchronic (one to three month) experi- ments in mice and rats. These tests revealed the range of dose levels that produced central nervous system depression and adverse hepatic (liver) effects. These effects. are believed to have a minimum threshold below which they will not occur. Environmental sampling data from industrial operations in which SOLVALL was heavily used indicate human exposure levels, at least 1,000 times lower than the minimum level resulting in central nervous system or hepatic toxicity in laboratory animals. It was therefore concluded that workers in such operations, believed to be the most highly exposed human population, were not at risk of developing adverse health effects. EPA’s Carcinogen Assessment Group (CAG) recently reviewed the available data on the carcinogenic potential of SOLVALL. One item was a cancer bioassay conducted in 1970 by the National Cancer Institute that was considered to be of adequate quality. In this lifetime inhalation exposure study, statistically signif- icant increases of kidney tumors were observed in male and female mice administered 150 or 300 mg/kg/day SOLVALL. In addition, statistically significant increases in liver tumors were observed in male and female mice and in female rats in this study. (No effect was observed in the 75 mg/kg/day group). Only one epidemiological study is available of solvent- distribution workers who were exposed to SOLVALL in addition to a number of other potentially carcinogenic chemicals. When these workers were compared with the adult male population of the United States, they were found to have a slightly increased rate of lung cancer, which was not statistically significant. The study did not control for smoking or for-the other chemical exposures of these workers. ------- 11—12 On the basis of these data, EP concluded that SOLVALL was a probable human carcinogen (Group 82). The animal data were con- sidered sufficient because there was a statistically significant increase in tumors in two species, with tumor incidence increas- ing as dose levels increased. The evidence from the human study was considered inadequate because while it showed evidence of an association, it did not exclude chance, bias, or confounding factors. A causal interpretation of these results in humans was therefore not judged credible. ------- 11—13 Issues to Be Considered On the basis of your evaluation of DNC, what questions would you ask the toxicologists about the hazards of SOLVALL? ------- III. HUMAN EXPOSURE EVALUATION SOME PRINCIPLES FOR EXPOSURE EVALUATION • The purpose of exposure evaluation is to identify the magnitude of human exposure to DNC and SOLVALf , the frequency and duration of that exposure, and the routes by which humans are exposed. It may also be useful to identify the number of exposed people along with other characteristics of the exposed population (e.g., age, sex). • Exposure may be based on measurement of the amount of DNC and SOLVALL in various media (air, water) and knowledge of the amount of human intake of these media per unit of time (usually per day) under different conditions of activity. • Some individuals may be exposed by contact with several media. It is important to consider total intake from - all media in such situations. • Because only a limited number of samples of various • media can be taken for measurement, the representative— ness of measured values of environmental contaminants is always uncertain. If sampling is adequately planned, the degree to which data for a given medium are representative of that medium can usually be known. • Sometimes air and water concentrations of pollutants can be estimated by mathematical models. Although some of these models are known to be predictive in many cases, they are not thought to be reliable in all cases. • Standard average values and ranges for human intake of various media are available arid are generally used, unless data on specific agents indicate that such values are inappropriate. SITE DESCRIPTION As discussed earlier, the Electrobotics Company, which manu- factures parts for Bananachrome personal computers, operates a production facility in eastern Massachusetts. The facility ------- 111—2 employs 150 workers and includes a wastewater treatment plant and equalization lagoon (surface impoundment). (The site plan is illustrated in Figure 2.) Wastewater from the manufacturing plant operations contains high concentrations of DNC and SOLVALL and is periodically discharged to a flow equalization lagoon, which is located within 200 feet of the western facility boundary. There are rio other potential sources of DNC or SOLVALL releases in the surrounding area. Adjacent to this boundary is a residential area of 20 houses in which 80 individuals reside. The area between the Electrobotics facility and the threatened water supply wells to the east is undeveloped and contains few residences. There has been some discussion about building housing on the property, but rio definite plans exist at this time. The equalization lagoon maintains a regular flow to an on—site a jy , dsu&g. wastewater treatment facility. The equalization lagoon is maintained at an average fluid depth of 10 feet and measures 10,000 square feet. Figure 2 Slte Plan and Points of Compliance and Exposure — — ROAO — Electrobotics Property Boundary I I ti Residential Area 7,000 ft. Characterization Wells Threatened Water Supply Wells Direction of I Ground-Water Flow Point of Compliance N (Not to Scale) ------- 111—3 The equalization lagoon is underlaid by a compacted natural clay liner with a hydraulic conductivity (a measure of the rate at which it transmits water) of 3 x l0 cm/sec, which is designed to provide substantial containment of the wastewater. The uppermost geologic formation beneath the site is composed of approximately 75 feet of stratified glacial outwash, which con- sists of layered sand and gravel with some silt. These unconsol- idated sediments form a productive aquifer that is a source of potable water to the surrounding area. A well field is located approximately 1.5 miles to the east (downgradient) of the site and provides approximately 3 million gallons per day (mgd) of potable water to its 50,000 customers. The well field contains 40 separate wells, each pumping 75,000 gallons per day (gpd). The hydraulic conductivity of the aquifer has been estimated at 40 feet per day, based on field tests in the area. However, there may be thin continuous layers of significantly lower or higher hydraulic conductivity within the glacial outwash. The water table is located about 10 feet below the bottom of the lagoon at the site. The hydraulic gradient (a measure of the slope of the water table), based on the best available potentio— metric head measurements, is estimated at 0.005 feet/feet toward the community well field. The site is located in a humid area. The wind direction is seasonal; however-, according to a wind rose from a local ineteoro— logical station (which describes wind direction and frequency), the wind blows west, across the site toward the 20 homes adjacent to the lagoon approximately 30 percent of the time. The owners of Electrobotics conducted extensive analysis of the lagoon in preparing their Part B application. The nature of the materials stored in the lagoon and the meteorological and hydrogeological conditions in the vicinity have resulted in some concern by EPA about possible exposure of workers at the facility and residents in the vicinity to DNC and SOLVALL via inhalation of air, as well as possible exposure of the community via con- tamination of drinking water supplies. In an attempt to be responsive to the regulations and concerns of EPA staff, the owners undertook a program of air and ground—water monitoring. Measurements of concentrations of DNC and SOLVALL on the site are described in the next section. A ground-water monitor- ing program has been undertaken to comply with regulations promulgated under the Resource Conservation and Recovery Act (RCRA). Electrobotics is monitoring contaminants in ground water at the downgradient limit of the waste management unit, termed the “point of compliance.” The point of compliance in this case is at the eastern limit of the berm around the lagoon. Three monitoring wells have been installed at this point, and one well was installed upgrádient (see Figure 2). This conforms to the ------- 111—4 regulatory minimum. No DNC or SOLVALL was found in the sample from the upgradient well. The nearest downgradient property boundary is to the east, about 1,000 feet from the lagoon. Because DNC and SOLVALL have been detected at the downgrad- ient compliance—monitoring wells, a ground—water assessment has been initiated, and additional monitoring wells have been con- structed at the eastern property boundary to monitor chemical concentrations. Further downgradient, the public well field represents a point of potential human exposure. Three of the 40 wells are located directly downgradient of the lagoon. The relative locations of the point of compliance, eastern regulatory boundary, and actual point of exposure are shown in Figure 2. AVAILABLE INFORMATION ON DNC AND SOLVALL CONCENTRATIONS Measurements of DNC and SOLVALL concentrations in air along the western site boundary, outdoors at the facility adjacent to the lagoon, and indoors within the treatment plant have been made during- a single air—sampling program. The sampling program was conducted for a period of one week, during concentrations were measureftă E he western property boundary, adjacent. to the equalization lagoon and within the treatment plant. During the period of measurement, wind was blowing gen- erally west across the lagoon toward the residential area. The mean, standard deviation, and range of measured chemical concen— trations in air are shown in Table 4. Air measurements that were made concurrently inside the wastewater treatment plant did not detect measurable concentrations of either solvent. Within the treatment plant, all treatment units are closed and vapor— controlled to limit any fugitive air-emissions. Table 4 FIELD I€ASUREMENTS OF OFIC f1D S(1V LL C(NCENTRATIONS ONC Sit VALL Detection Standard Standard Nediup Location Level Mean Deviation Range - Mean Deviation Range Air Ineide treatment plant ( Lg/in 3 ] ND 1 — — pu I — — Air At weatern boundary of mite iug/m 3 1 11 4 2—17 153 10 130—160 Air l 1—site Iug/m 3 ] 47 20 30—120 700 22 650—1 000 Ground water Point of compliance (ug/l] 332 30 290—350 1,000 168 000—1,250 Ground water Eastern property boundary 90L 2 0012 Ground watar Municipal water supply ND’ — ND’ ‘Not detected at 1 ppb. 2 Trace concentretiona. below detection limit at 1 ppb. ------- 111—5 The ground water has been periodically sampled and tested for DNC and SOLVALL both at the point of compliance, the monitor- ing wells on the eastern boundary, and at the water treatment plant at the community water system, which draws water from the aquifer for the municipal supply. The concentrations of chemi- cals in monitoring wells at the point of compliance are indicated in Table 4. The concentrations detected at the compliance point indicate contamination of the aquifer. Trace concentrations of SOLVALL have also been detected at the eastern boundary in the most recent water tests, but at concentrations less than the method—detection limit of 1 part per billion (ppb or ig/l). The municipal water supply has been tested, and no detectable concen- trations of DNC or SOLVALL were found. However, the detection limit for both chemicals offered by available analytical methods for ground water is 1 ppb. Available information suggests that the chemical plume may not yet have migrated beyond the eastern property boundary, but without corrective action the concentrations are expected to increase in time in this vicinity, eventually affecting offsite ground water. A mathematical model of the contaminant plume movement in the aquifer has been constructed to estimate future concentra— tions in the aquifer. A summary of the modeled environmental concentrations in ground water is presented in Table 5. Appen- dix A contains the calculations and the associated assumptions that were applied in modeling these environmental concentra- tions. Table 5 SU*IARY OF MODELED CONCENTRATIONS OF ONC ND SOL VAIL IN GROUND WATER Location DNC (ag/i) SOLVALL (ag/i ) Eastern property boundary 275 770 Actual paint of exposure 5.5 9.3 ------- 111—6 CALCULATIONS OF HUMAN EXPOSURE The environmental concentrations summarized in Tables 4 and 5 are the starting point for the calculation of estimates of human exposure to DNC and SOLVALL. The medium in which a substance is present will determine the potential route of human exposure. For example, substances present in water may be ingested. Contaminated water may also lead to inhalation exposure when water is used for cooking or showering, although this exposure route is likely to be minor unless the contaminant is present at high concentrations in the water and is highly volatile. Exposure by inhalation of contam- inated water or by the derma]. route during bathing or showering vlas considered insignificant because of the low concentrations of these substances in the municipal water supply. Doses resulting from these exposure routes were therefore not calculated. Sub- stances present in air will be inhaled. Finally, substances present in soil may be ingested, absorbed through the skin, inhaled, or taken up by plants with human exposure resulting if these plants are used as food or if these plants are fed to livestock, the various products of which are used as food. Human exposure to contaminated soil, however, was considered insignificant at this site because soils potentially contaminated by ground water are located at least 10 feet below the land sur- f ace. Consequently, doses resulting from potential human expo- sure to soil were not calculated. In order to estimate human exposure, or dose of the consti- tuent from each contaminated medium, certain data and assumptions were applied. These data and assumptions relate to the extent and frequency of human contact with these media and the degree of absorption of chemicals for each route of exposure. When the constituent is ingested, a certain fraction will be absorbed through the gastrointestinal wall; when it is inhaled, a certain fraction will be absorbed from the lungs. The method of dose calculation depends on the route of expo- sure. Certain standard values have been developed to estimate contact with and intake of certain media. For drinking water, the EPA and other scientific groups generally assume that adults drink 2 liters of water per day. It is also generally assumed that the average adult inhales 23 cubic meters Cm 3 ) of air per day. It should be noted, however, that the data and assumptions required for estimation of dose from most other routes of exposure are not as readily standardized. ------- 111—7 The assumptions and calculations required to estimate DNC dose to the 80 neighboring residents resulting from inhalation of DNC—contaminated air are presented below to illustrate the ele- ments of dose estimation. A complete set of calculations and associated assumptions for both chemicals and the critical pathways is shown in Appendix B. Inhalation of DNC—Containinated Air by Neighboring Residents Assumptions: • An adult inhales 23 m 3 /day of air. • Based on wind direction analysis, the duration of exposure is 30 percent of the time on an annual average basis. • The body weight of an adult is 70 kg. • The inhalation absorption factor for DNC is 0.75. • The adult lives in the home throughout his lifetime. Calculations: 0.011 mg (average DNC air concentration at boundary) m 3 23m 3 1 x x x 0.3 (human intake factor) day 70 kg x 0.75 (inhalation absorption factor) = 8.1 x l0 mg/kg/day Table 6 presents a summary of all the exposure calculations performed to estimate doses from the contaminated media to the identified exposed population groups. ------- 111—8 Table 6 SUI4IARY OF RESULTS OF EXPOSURE CALCULATIO S N anber Methtai DIC (ng/kqjday) SOLVALL (mq/kg/day) of Persons Exposed Air, neighboring residents 8.1 * iO 1.1 x io.2 80 Air, workers 1.2 x 10 1.8 x 10.2 150 Ground water, point of compliance 9.5 * 1O 2.9 x ia. 2 0 Ground water, eastern property boundary 7.8 x i0 2.2 x LO.. . 2 0 Ground water, public well field 1.6 * 10 2.7 * 10 50,000 Remarks on Exposure Data 1. Ground water directly beneath and adjacent to the regulated unit has been sampled and analyzed for the presence of DNC and SOLV LL. Concentrations that have been measured have remained relatively constant for the past two years (four samplings) and indicate that both these chemicals have leaked from the storage lagoon into the underlying aquifer. Ground water moves toward a well field that provides drink- ing water for a community of 50,000 people. DNC and SOLVALL have not been detected in the water supply but may be pres- ent at concentrations below the current method detection limits, which are 1 ppb ( .tg/l) for both chemicals. 2. Both DNC and SOLVALL are unstable in the environment and will decompose by biodegradation in shallow aquifer systems. The decomposition of both chemicals in aquifers has been demonstrated in aquifer restoration programs at other sites, but the rates of decomposition are variable and have only been quantified in laboratory experiments. The rate of decomposition at concentrations less than 10 ppb is uncertain. 3. Trace concentrations (less than 1 ppb) of chemicals have been detected at the eastern facility boundary. A mathemat- ical model has been used to predict future concentrations of DNC and SOLVALL in ground water downgradient of the regu- lated unit. The model prediction indicates that both chemi- cals will migrate in ground water and degrade at a slow rate. On the basis of the model prediction, concentrations are expected to increase at the eastern property boundary if no corrective action is taken. The degradation products are not believed to be carcinogenic. ------- 111—9 4. Estimates of future concentrations of both chemicals in drinking water are based on analytical predictions from a ground—water model. Appropriate adjustments regarding ground—water dilution were made to estimate concentrations at the tap. 5. Estimates of current exposure in air were based on analyti- cal results obtained from measuring the air concentrations in the facility and at the boundary over a single seven—day period in April 1986. Twenty—four—hour composite samples were taken each day over the sampling period. No other air data are available. ------- 111—10 Issues to Be Considered 1. The summary exposure data presented in Table 6 include some current exposure estimates and some future expo- sure estimates. Models are used to predict concentra- tions of DNC and SOLVALL at some point in the future, assuming that releases of the substance from the lagoon’ occur at the current rate. Should the two types of exposure data (current and future) be treated the same for purposes of characterizing human risk? How should distinctions be reported? 2. Should the ground—water concentration estimates that are based in part on modeling of a degradation process be used at all? How should this be decided? If modeled concentrations are not used, how should the analyses of chemicals that are reported as “below detection limit” or “not detected° in the water supply system be represented in exposure estimates? 3. Is the mean concentration in the various media the appropriate summary statistic to use to characterize human exposure? Should the tipper range or statistical upper confidence limit be used as an alternative? 4. Are the various assumptions about human intake and average exposure to various media valid? Should others be substituted or added? 5. Should other routes of exposure to the various contami- nated media have been considered? 6. Should the total exposure and risk for the regulated unit be represented by the sum of all incremental risks for each chemical/pathway? 7. Was it appropriate to model the exposure based on an adult population? Should other populations—at-risk have been considered? 8. Should exposure and risk to workers at the facility be considered in the same context as residents in the surrounding community? ------- 111—11 Some Possible Conclusions About Human Exposure to DNC and SOLVALL Which of the following conclusions best characterizes the information you have seen? 1. None of the exposure estimates is adequate for use in risk assessment. The risk assessment should describe exposure in qualitative terms only. Such a qualitative description is appropriate and adequate for character- izing risk, which also can be done in qualitative terms only. 2. The exposure estimates presented in Table 6 for drink- ing water and air are reliable and can be used for risk assessment. However, no risks should be assessed until appropriate data are obtained from other pathways of exposure. 3. Although the exposure estimates in Table 6 are based on different data and assumptions, they are all adequate and sufficient for assessing risks. The risk manager should be made aware of the uncertainties in each of the data sets, but a quantitative risk should be devel- oped. 4. In addition to Conclusion 2, it should be noted that all the exposures from various media should be added for those people exposed to all sources of DNC and SOLVALL. 5. Other (formulate your own). ------- IV. DOSE-RESPONSE EVALUATION Normally, you would not be involved in reviewing dose- response information; the results generally would be pro- vided to you. As discussed before, however, we hope that by having you evaluate this information and address the key issues here, you will be able to better use the dose— response information you will receive. We have given you the EPA approach to evaluating dose- response relationships. In addition, we have identified alternative approaches so that you develop an understanding of how others may perceive the issue. ------- IV-2 DOSE-RESPONSE EVALUATION: THE GENERAL-PROBLEM AND PRINCIPLES GUIDING APPROACHES TO ITS SOLUTION Because of the relative complexity of dose—response evalua- tion, the following discussion is substituted for a statement of key principles. Animal data showing that DNC and SOLVALL are carcinogenic were obtained in the high—exposure region of the dose—response curve. Thus, animal exposures were in the 30—to—300—unit range (see Table 1 for DNC), and these produced measurable risks in the range of 10 to 50 percent (see Table 2 for DNC). Predictions of human exposure, as discussed in Section III, are at much lower levels. What can be said about risks in the range of human exposure? At least three general approaches to this problem have been proposed by various experts. Approach 1 Based on general theories of how carcinogens act to produce cancer (largely derived from experimental studies and epidemi- ological data), all finite exposure levels will produce a finite risk. The magnitude of the risk will decline as the magnitude of exposure declines (clear even in the animal data). 1 If the quantitative relationship between exposure and risk were known for all exposures, the risks to rodents exposed at very low levels could be predicted from the measured exposure— risk data. The risks to humans could be predicted at these very low levels if the relationship between rodent and human suscepti- bilities were known. Although these relationships cannot be known with accuracy, a plausible upper limit on human risk can be predicted with sufficient accuracy to be used as a guide to mak- ing risk decisions. Actual human risk is not likely to exceed the upper limit, and it may be less. This is the approach generally adopted by EPA in evaluating the risk associated with low—level exposure to carcinogens. 1 These two sentences are the proper formulation of the “no— threshold” concept. It does not mean that all finite exposures will cause cancer; rather, it means that all finite exposures will increase the probability (risk) that cancer will occur. ------- IV-3 Approach 2 The quantitative relationships between high—exposure and low—exposure risks in rodents and between rodent and human risk are not known with sufficient reliability to be used in risk assessment. Moreover, there is no reliable theory on which one can conclude with assurance that low—level human exposure (i.e., exposure below the range producing detectable risks) poses any risk at al] .. As with other toxic effects, carcinogenicity will not be initiated within an individual until a minimum threshold of exposure is exceeded. In such circumstances, the only reason- able course is to report the magnitude of the margin—of—exposure (MOE) by which humans are protected. MOE is the maximum amount of exposure producing no measurable tumorigenic response in ani- mals divided by the actual amount of human exposure. MOE gives the risk manager adequate information on which to decide whether exposures must be reduced or eliminated to provide human protec- tion. A relatively large MOE is desirable because it is likely that the threshold for the entire human population is lower than that observed in small groups-of experimental animals. This approach is generally applied when evaluating the potential risk of most noncarcinogenic effects. Approach 3 Although there is adequate theory and some evidence to per- mit the conclusion that humans are at finite risk at all finite exposure levels, there is insufficient knowledge to allow predic- tion of the risks in quantitative terms. The risk assessor should simply attempt to describe risks qualitatively, perhaps coupling this description with some information on the potency of the compound and the magnitude of human exposure. This type of presentation is adequate for the risk manager, who should nřt be concerned with the quantitative magnitude of risk in any case. - * * * * * Each of these views, and perhaps others as well, has some merit. The first approach is now used by most federal public health and regulatory agencies, including EPA. These agencies emphasize that the predicted numerical risks are not known to be accurate, but because of the nature of the models used to predict them, they are likely to be upper—bound estimates of human risk . An upper—bound estimate is one that is not likely to be lower than the true risk and is likely to exceed the true risk (which could be zero). . - ------- IV-4 For this exercise, we will estimate low—exposure risks using the model currently used by EPA. A model is a mathematical for- mula that describes the relationships between various measures. Two models are needed to predict low—exposure risics: • A high to low exposure extrapolation model is needed to predict low—exposure risks to rodents from the measured high—exposure, high-risk data (see Table 2 for DNC). EPA currently uses a “linearized multistage model” for this purpose. This model is based on general (not chemical—specific), widely held theories on the biolog- ical processes underlying carcinogenesis. Application of the model to the rodent exposure risk data produces an estimate of the lifetime risk for each unit of exposure in the low—exposure region. This is called the unit cancer risk. The linearized model is used to ensure that the unit cancer risk is an upper—bound estimate of risk. • An interspecies extrapolation model is used to extra- polate from rodent unit risks to human unit risks. EPA assumes that rodents and humans are at equal risk at the same exposure measured in milligrams of carcinogen per square meter of body surface area per day. Inter- species extrapolation models are commonly called “scaling factors” because they are used to scale doses between species. EPA’S selection of these models is based on the agency’s view that they are the best supported for purposes of deriving an upper-bound estimate of risk. Alternative models are available for both these forms of extrapolation, and several are equally plausible. In most cases, but not always, use of plausible alternative models will yield lower estimates of risk than those predicted by the two described here. Differences can sometimes be very large, but are generally relatively small when the models are limited to those that are linear at low exposures. Further discussions of various models and their plausibility can be found in the handout “Principles of Risk Assessment: A Nontechnical Review.” APPROACH TAKEN FOR THIS EXERCISE In this exercise we determine the upper—bound estimate of unit cancer risks predicted for DNC and SOLVALL using the models currently preferred by EPA. The effect of using alternative, ------- IV-5 plausible models is also described, as are the animal data that pertain to the second or third approaches described above for dose—response evaluation (i.e., the MOE or qualitative approaches). Estimates of Upper—Bound, Lifetime Unit Cancer Risks Using Current EPA Models Application of the EPA models for high—to—low—dose and interspecies extrapolation to the measured animal cancer data for DNC (Table 2) and SOLVALL yields the results shown in Table 7. The result of such extrapolations is the unit cancer risk. In this exercise, one unit of exposure is equal to one milligram (mg) of the carcinogen per kilogram (kg) body weight of the animal (or human) per day throughout the lifetime of the animal (or human). Thus, if the unit cancer risk for DNC is 0.0897 (8.97 x 10—2), the risk of developing cancer from 1 mg/kg/day of lifetime exposure to DNC is 8.97 x 10—2 (a probability of about 9 in 100). Table 7 UPPER—BOUND ESTIMATES ON LIFETIME UNIT CANCER RISKS PREDICTED FROM APPLICATION OF EPA’S PREFERRED MODELS TO ONC AND SOLVALL TUMOR DATA Species, Sex Route of Exposure Tumor Site Unit Cancer Risk 1 DNC (based on Table 2) Rat, male Inhalation Lung 0.0186 (1.86x10 2 ) Rat, male Inhalation Spleen 0.0126 (1.26x10 2 ) Rat, male Inhalation Liver 0.0168 (1.68x10 2 ) Rat, male Gavage Stomach 0.0054 (5.4x11r 3 ) Rat, female Cavage Stomach 0.0054 (5.4x1O 3 ) Rat, male Gavage Liver 0.0120 (1.2x10 2 ) Rat, male Gavage Spleen 0.0228 (2.28x1O 2 ) Mouse, male Gavage Liver 0.0897 (8.97x1O 2 ) Mouse, male Gavage Stomach 0.0096 (9.6x10 3 ) SOIVALL (as provided by EPA Carginogen Assessment Group) Mouse, male Inhalation Pooled tumors 2 0.0018 (1.8x10 3 ) 1 Risk for an average daily lifetime exposure of one unit. Units are the same as those used earlier for describing the animal exposure (Table 1 for ONC), such that one unit = one milligram per kilogram body weic t per day. Risk is obtained from unit risk by multiplying the latter by the actual number of units of human exposure. For a given exposure, the hi er the unit risk, the higher the risk. 2 The percent of animals with one or more tumor sites showing signif i— cantly elevated tumor incidence. ------- IV-6 This relationship is presented graphically in Figure 3. The horizontal axis is the dose of DNC measured in mg/kg/day. The vertical axis is the response and is a probability. Figure 3 Dose—Response Relationship for DNC Response 0897 (unit cancer risk) To obtain the risk from other levels of exposure, the unit cancer risk is multiplied by the number of units of exposure. Thus, if a person were exposed to 0.003 units of DNC throughout his lifetime (0.003 mg/kg/day), the risk would be 0.003 units times 0.0897/unit, or 0.00027 (2.7 x l0 ), a probability of 2.7 cases arising in 100,000 individuals exposed at this level. Estimates of Lifetime Unit Cancer Risks Using Other Models Application of other models for high-to—low—dose extrapola- tion yields unit risks equal to or slightly lower than (less than fivefold) those in Table 7, as long as the other models incor- porate the concept that risk increases in direct proportion to exposure in the low—exposure region (linear models). Use of the most plausible alternative interspecies extrapolation model (that generally used by FDA) yields unit risks sixfold (for rats) and thirteenfold (for mice) lower than those predicted in Table 7. Thus, alternative models predict unit risks about 30 to 65 times lower than those predicted using the EPA models. Adoption of certain nonlinear models for high—to—low—dose extrapolation predicts risks about 1,000 to 10,000 times lower than those predicted by use of the EPA model. The nonlinear models are not used by agencies charged with protecting human health, but cannot be rejected on purely scientific grounds. Dose (m9lkgiday) ------- IV-7 Dose—Response Evaluation not Involving Formal Extrapolation For those who believe formal extrapolation beyond the measurable dose—response data should not be performed, it is important to identify the exposures at which DNC and SOLVALL produce tumors and those at which no tumor excess is found (the “no observed effect level” or NOEL). Table 8 identifies NOELs from data on DNC in Table 2. Note that this is not the approach generally applied by EPA for carcinogenic substances. (The Carcinogen Assessment Group consequently did not report NOELs for SOLVALL). Table 8 NO—OBSERVED EFFECT LEVELS (NOELe) FOR CHRORIC EXPOSURE TO DNC 1 Study Group Sex Ti.sior - NOEL ONC (based an Table 2) Rat, inhalation Male Lung 30 Rat, inhalation Male Spleen 30 Rat, inhalation Male Liver 30 Ret, gevage Male Stomach 50 Rat, gavage Female Stomach 50 Rat, gavage Male Liver 50 Rat, gavage Male Spleen None found Mouse, gevage Male Liver None found Mouse, gevage Male Stomach 60 1 tjnits are identical to those in Tables 1 end 2. “None found” means that a measurable excess of tumors was found at both levels of exposure used in the experiment. ------- IV-8 Issues to Be Considered 1. If explicit estimates of unit risks are made, should only EPA’S currently preferred models be used? Should the results of applying other models also be displayed? 2. Which species/sex/tumor site data from Table 7 should be used for unit risk assessment? All, shown individ- ually as in Table 7? Only the data set yielding the highest unit risk? A sum of all? Other? 3. Should the DNC stomach tumor data set be rejected because there is no exact anatomical counterpart in humans? 4. How should the uncertainties in use of models be described? 5. Are the observed NOELs true “no-effect” levels? Could they simply reflect the fact that in experiments with relatively small numbers of animals, the failure to observe a statistically significant increase of tumors is, an artifact of the experimental design, and not a true absence of biological effect? How should this uncertainty, if it is real, be taken into account? 6. EPA has adopted the first approach, using unit cancer risks, to extrapolate from the high doses used in the animal studies to the low doses at which humans are exposed. How might you respond to someone who argues for one of the other approaches? ------- IV- 9 Some Possible Conclusions About Dose—Response Evaluation Which of the following conclusions best characterizes the information you have seen? 1. The unit cancer risks listed in Table 7 are true upper— bound estimates. The true unit risk is not likely to exceed those listed, may be lower, and could be zero. 2. The same as the first conclusion, but add: The use of alternative, plausible models yields unit risks about 10 to 100 times lower than those in Table 7. 3. Unit risks should be reported for all plausible models, and the full range of estimates should be reported without bias. 4. There is no justification for calculating and reporting unit risks. What is critical for understanding the public health importance of low—level exposure to DNC (and SOLVALL) is the margin of exposure (MOE). Estima- tion of the MOE is based on the NOELs for its carcino- genic effects; these figures are reported in Table 8. 5. Neither unit cancer risks nor NOELs are reliable indi- cators of human risk, and neither should be considered for risk assessment. Dose—response relations for the human population are not known for DNC or SOLVALL; risk should be described in qualitative terms only. 6. Other (formulate your own conclusion). ------- V. RISK CHARACTERIZATION PURPOSE In the last step of risk assessment, the information col- lected and analyzed in the first three steps is integrated to characterize the excess risk to humans. In line with the alter- native approaches for describing dose—response relationships, at least three approaches to this step can be taken: 1. Provide an explicit numerical estimate of excess lifetime risk for each population group by multiplying the unit risk times the number of units of exposure experienced by each group: Excess lifetime = (unit cancer risic) x (units of exposure) risk In this equation, excess risk is unitless——it is a probabil- ity. 2. Estimate the margin of exposure (MOE) for each group by dividing the NOEL by the exposure experienced by that group. 3. Describe risks qualitatively for each population group. Risk characterization might also include some combination of all three approaches, along with a description of their relative merits. It is also essential that the statistical and biological uncertainties in estimating the extent of health effects be described in this step. Table 9 presents the excess lifetime risks for each popula- tion group using data from Tables 6 and 7. These risks are based on the highest unit cancer risk for DNC and the unit risk deter- mined by the Carcinogen Assessment Group for SOLVALL (Table 7). If other unit risk figures from Table 7 had been used, excess risks would be somewhat lower. And if unit cancer risks derived from other dose response models had been used, the excess risics shown in Table 9 would be 10 to 100 times lower. The risks in Table 9 are thought to be upper—bound lifetime risks. Table 9 also reports the MOE for each group. The MOE is not an expression of risk, but the difference between the actual ------- V-2 human exposure and the highest exposure causing no observable carcinogenic effect in the most sensitive animal species. (Use of the MOE approach is not the policy of the EPA. These results are presented, however, because this approach may be applied by non—agency risk assessors.) - Table 9 UPPER—BOUND ESTIMATES (F EXCESS LIFETDC HUMAN RI5K AND MARGINS OF EXPOSU FR 4 EXPOSU To DETECTED DEMICALS 1 RIsk obtained by multiplying the highest unit cwieer risk Figure frow Table 7 (0.0897 for ONC and 0.0018 for SCLVALL) by the unite of exposure from fable 6. 2 Tota l risk obtained by udding risks from air and aster exposures. 3 1U For ONC obtained by dividing the lowest NOEL (30 mg/kg/day) from Table 0 by the mis of ONC air and water exposures from Table 6. For SOLVALL, HOE obtained by dividing 75 mg/kg/day (the loweet NOEL) by the mum of S .VALL air and water exposures from Table 6. 4 0btsined by multiplying total excess risk by thu ustisatud riuaber of people exposed. Exposed Population Upper —Oound Estimate of Exceas Excasa Chesics.l Risk from Air’ Excess Risk Cancer Cease from Water 1 Total Excess Risk 2 Over a Lifetime 4 General population (50,000) Nearby residents (80) Workers (150) Risks Based on Water Consumption at Pi 1ic Well Field DNC 501 VA IL DNC 7.3 x LO S01VALL 2.0 x 1O 9.3 a l0 ONC S aL VAIL 1.4 * 1.4 a l0 1.4 x 4.9 a i0 1.4 a 1.4 x l0 4.9 a 10 1.4 x 10 1.4 a 1O 4.9 a l0 1.4 a 8.1 a 2.0 a 1.1 * ia— 4 1.2 a 3.2 a l0 1.5 a 10 1.1 a 10-4 3.2 a 10 1.4 a 10 1.9 a 3.0 a 10 l0 0.7 3.1 a 6.8 a 10 0.009 2.2 a 4.0 a l0 10 0.02 Risks Based on Water Consumption at Eastern Property Bowidary General population (50,000) ONC SO t VALL Nearby residents (80) DM0 501 VAIL Workers (150) DM0 50 1. VAIL 7.0 a 4.0 10 10 7.0 a 4.0 • 10 la_S 3.8 a 3.5 a i ’ 10 ’ 7.4 a 10 7.4 a 10 1.3 a 2.0 a l0 7.0 x 4.0 a 10 7.7 a 6.0 a 10 3.5 a 2.3 a l0 IO’ 9.3 a l0 7.4 x ]0 0.3 a 10-4 1.1 3.2 a i — IO 7.0 4.0 a i — 8.1 a 7.2 a ia— l0 s. a 1.9 a i0 jQ3 1.4a1 0 7.4xl O 8.Oxl O Risks Based on Water Consumption at Point of 37 0. 0 0.1 General population (50,000) DNC SOLVALL 8.5 x 5.2 a 10 l0 8.5 x 5.2 a l0 i0 5.2 a 2.8 a 10’ l0 9.0 a 10 9.0 a iO 85 Nearby residents (00) ONC SOLVAIL 7.3 a 2.0 a i0 lo 8.5 a 5.2 x 10 lO 9.2 a 7.2 a i — i0 2.9 a 1.9 a l0 io 9.3 a 10 9.0 a 10 9.9 a 10 0.08 Workers (150) DM0 S OLYA I.L 1.1 a 3.2 a l0 lO 8.5 * 5.2 a 10 9.6 a 0.4 a l0 l0 2.0 a 1.6 a 10’ 1O 1.4 x l0 9.0 a i0 4 1.0 * 10 0.15 ------- V- 3 Issues to Be Considered 1. Are the results reported in Table 9 an adequate characteriza- tion of DNC and SOLVALL risks? What else should be added? 2. Risks are presented at three points: the point of compliance, the eastern property boundary, and the public well field. What are the advantages and disadvantages of using each? Which one do you feel best represents the risks? 3. Should risks derived using other unit cancer risks reported in Table 7 and unit risks obtained using alternative models also be discussed? How would you. present these uncertainties? 4. The risks, MOE, and number of cases reported in Table 9 depend on the assumption that the number of people exposed and their level of exposure will remain constant over a lifetime. Is this a plausible assumption? Can alternative assumptions be used? 5. Is it important to distinguish routes of exposure? Should unit risks obtained from the gavage data be used for population groups exposed by inhalation? Should gavage data be used at all? 6. The risks from DNC and SOLVALL are added together in Table 9. Does this seem reasonable? 7. Is it appropriate to estimate the number of cancer cases by multiplying risk times population size (last column of Table 9)? Which is more important——risk to an individual, or risk to a population? 8. Do you believe that animal data obtained from continuous, hf e— time exposure should not be used to characterize the risk of people exposed by certain routes intermittently, for a rela- tively small fraction of their lifetime? ------- V- 4 Some Possible Conclusions About DNC and SOLVALL Risks Which of the following best characterizes the information you have seen? 1. Upper—bound excess risks to humans exposed to DNC and SOLVALL are those reported in Table 9. Although risks obtained from the use of other models are lower, the risks could be as high as those reported in the table. There is no evidence that a threshold dose exists for DNC or SOLVALL, so the MOE estimates have little meaning. 2. Same as Conclusion 1, except restrict estimates of excess risks for inhalation exposure to unit risks estimated from inhalation data, and restrict risks for ingestion to gavage data. 3. The excess risks shown in Table 9 as well as those obtained from all other plausible models and all the various tumor site data should be reported, and all estimates should be given equal weight. Such a presentation affords the decision maker a view of the uncertainty in the estimated risks. 4. Upper—bound estimates of excess lifetime risks to humans are those reported in Table 9. Use of all other animal data sets and alternative risk models used by some other agencies would result in prediction of lower risks, perhaps up to 65 times lower. These risks are conditional on the assumption that DNC and SOLVALL are probable human carcinogens, based on observations of carcinogenicity in two species of experimental animals. Uncertainties in the exposure and population estimates are those described in the Exposure Assess- ment section. 5. DNC and SOLVALL are probable human carcinogens, based on observations of carcinogenicity in two animal species. Exposures needed to produce animal carcinogenicity are many times higher than those to which humans are exposed. The margins of exposure by which humans are protected are shown in Table 9. Because a NOEL has not been identified for all the various carcinogenic end points, a greater than usual MOE should be employed to protect human beings. ------- V- 5 Some Possible Conclusions About DNC and SOLVALL Risks (continued) 6. DNC is a probable human carcinogen, based on observations of carcinogenicity in two species of experimental animals. SOLS/ATJL is a probable human carcinogen, based on strong evidence in two species of experimental animals and suggestive evidence from observations in exposed humans. Humane are exposed to these chemicals through air and water. In general, large numbers of people are exposed continuously to very low levels of DNC and SOLVALL in drinking water, and a few groups are exposed to relatively high levels in air, some continuously, others intermittently. The individual risk in the general (larger) population is probably low, and this translates to a very small number of cancer cases. The individual risk to nearby residents is moderate, and this also translates to a small number (<1) of excess cancer cases. The individual risk to workers at the plant is slightly higher than that of nearby residents, although because of the relatively small number of workers, this again translates to less than one excess cancer case expected in this group. 7. Other? Some combination of the others? ------- GLOSSARY Acceptable daily intake (ADI) . Estimate of the largest amount of chemical to which a person can be exposed daily that is not anticipated to result in adverse effects (usually expressed in mg/kg/day). Carcinogen . A substance that increases the risk of cancer. Control animals . Animals that receive identical treatment as test animals, except exposure to DNC, for the purpose of observ- ing the natural or background rate of cancer in that type of animal. Dose . Measurement of the amount received by the subject, whether human or animal. Dose—response evaluation . A component of risk assessment that describes the quantitative relationship between the amount of exposure to a substance and the extent of toxic injury or disease. Dose—response relationship . The quantitative relationship between the amount of exposure to a substance and the extent of toxic injury produced. - Epidemiological study . Study of human populations to identify causes of disease. Such studies often compare the health status of a group of persons who have been exposed to a suspect agent with that of a comparable unexposed group. Exposure . To be accessible to the influence of a chemical or chemical action. Extrapolation . The estimation of a value beyond the known range on the basis of certain variables within the known range, from which the estimated value is assumed to follow. Gavage . Type of exposure in which a substance is administered to an animal through a stomach tube. Hazard evaluation . A component of risk assessment that involves gathering and evaluating data on the types of health injury or disease (e.g., cancer) that may be produced by a chemical and on the conditions of exposure under which injury or disease is produced. ------- 2 High—to—low—dose extrapolation . The process of predicting low- exposure risks to rodents from the measured high—exposure, high—risk data. Human exposure evaluation . A component of risk assessment that involves describing the nature and size of the population exposed to a substance and the magnitude and duration of exposure. The evaluation could concern past exposures, current exposures, or anticipated exposures. Human health risk . The likelihood (or probability) that a given exposure or series of exposures may have or will damage the health of individuals experiencing the exposures. Incidence of tumors . Percentage of animals with tumors. Interspecies extrapolation model . Model used to extrapolate from results observed in laboratory animals to humans. Linearized multistage model . Derivation of the multistage model, where the data are assumed to be linear at low doses. Margin of safety (MOS) . Maximum amount of exposure producing no measurable adverse effect in animals (or studied humans) divided by the actual amount of human exposure in a population. - Microgram (JLg) . One-millionth of a gram (1 ,zg = 3.5 x 10—8 oz. = 0.000000035 oz.). - Milligram (i g) . One-thousandth of a gram (1 mg = 3.5 x i0 oz. = 0.000035 oz.). Multistaqe model . Mathematical model based on the multistage theory of the carcinogenic process, which yields risk estimates either equal to or less than the one—hit model. Neoplasm . An abnormal growth of tissue, as a tumor. No—observed-effect level (NOEL) . Dose level at which no effects are noted. One—hit model . Mathematical model based on the biological theory that a single “hit” of some minimum critical amount of a carcino- gen at a cellular target——namely DNA-—can initiate an irrevers— - ible series of events, eventually leading to a tumor. Potency . Amount of material necessary to produce a given level of a deleterious effect. ------- 3 . Parts per billion. Parts per million. Risk . Probability of injury, disease, or death under specific circumstances. Risk assessment . The scientific activity of evaluating the toxic properties of a chemical and the conditions of human exposure to it both to ascertain the likelihood that exposed humans will be adversely affected, and to characterize the nature of bhe effects they may experience. Risk characterization . Final component of risk assessment that involves integration of the data and analysis involved in hazard evaluation, dose-response evaluation, and human exposure evalua- tion to determine the likelihood that humans will experience any of the various forms of toxicity associated with a substance. Risk management . Decisions about whether an assessed risk is sufficiently high to present a public health concern and about the appropriate means for control of a risk judged to be significant. Route of exposure . Method by which the chemical is introduced into the biological organism. Safe . Condition of exposure under which there is a “practical certainty” that no harm will result in exposed individuals. Scientifically plausible . An approach or concept having substan- tial scientific support but without complete empirical verification. Statistically significant . The difference in tumor incidence between the treated and control animals that is probably not due to chance. Systemic effects . Effects observed at sites distant from the entry point of a chemical due to its absorption and distribution into the body. Threshold dose . The dose that has to be exceeded to produce a toxic response. Total dose . Sum of doses received by all routes of exposure. ------- 4 Unit cancer risk . Estimate of the lifetime risk caused by each unit of exposure in the low—exposure region. Upper—bound estimate . Estimate not likely to be lower than the true risk. ------- Appendix A GROUND-WATER MODELING CALCULATIONS AND ASSOCIATED ASSUMPTIONS Predicted Concentration of DNC in Ground Water at Eastern Property Boundary and Public Well Field • Concentration in lagoon: 30 ppm (mg/i) • Concentration measured at compliance point: 332 ppb g/l) • Half—life of DNC in aquifer: 10 years = 3,650 days ——Degradation rate of DNC = 0.693/half—life = 1.9 x l0 day 1 Distance from source to property boundary: 1,000 feet • Travel time (T) from the source to property boundary: (distance traveled] x (effective porosity] T= (hydraulic conductivity] x (hydraulic gradient] 1,000ft x 0.2 = ___________________ = 1,000 days 40 feet/day x 0.005 • Steady-state concentration at property boundary: (332 ppb] x exp ((—1.9 x day 1 ) x 1,000 days] = 275 ppo • Distance from source to public well field: 1.5 miles = 8,000 feet ------- A-2 • Travel time from the source to three threatened wells: 8,000 x 0.2 ___________________ = 8,000 days 40 feet/day x 0.005 • Steady—state concentration at the three threatened wells: [ 332 ppb] x exp ((—1.9 x i0 day 1 ) x 8,000 days] = 73 ppb • Steady—state concentration at actual point of exposure (drinking water supply): The three threatened wells provide 7.5 percent of total drinking water supply; therefore, the concentration at the actual point of exposure (the drinking water supply): 0.075 x 73 ppb = 5.5 ppb Predicted Concentration of SOLVALL in Ground Water at Eastern Property Boundary and Public Well Field • Source concentration in lagoon: 50 ppm • Concentration measured at compliance point: 1 ppm • Half-life of SOLVALL in aquifer: 7.3 years = 2,660 days -—Degradation rate of SOLVALL = 0.693/Half—life = 2.61 x i0 day 1 • Travel time from point of compliance to property boundary = 1,000 days • Steady—state concentration at property boundary: (1 ppm] x exp [ (—2.61 x l0 day 1 ) x (1,000 days)] = 770 ppb ------- A— 3 • Travel time from point of compliance to three threatened wells = 8,000 days • Steady—state concentration at three threatened wells: (1 ppm] x exp ((—2.61 x i0 day -) (8,000 days)] = 124 ppb • Steady—state concentration at public well field (drinking water supply): The three threatened wells procride 7.5 percent of drinking water supply; therefore, the concentration at the actual point of exposure (in the water supply): 0.075 x 124 ppb = 9.3 ppb ------- Appendix B EUMAN DOSE CALCULATIONS AND ASSOCIATED ASSUMPTIONS Inhalation of DNC—Contaminated Air by Neighboring Residents Assumptions: • An adult inhales 23 m 3 /day of air. • Based on wind direction analysis, the duration of exposure is 30 percent of time on an annual average basis. • The body weight of an adult is 70 kg. • The inhalation absorption factor for DNC is 0.75. • The adult lives in the home throughout his lifetime. Calculations: 0.011 mg (average DNC air concentration at boundary) m 3 23m 3 1 x x x 0.3 (human intake factor) day 70 kg x 0.75 (inhalation absorption factor) - = 8.1 x l0 mg/kg/day Inhalation of DNC—Contaminated Air by Workers Assumptions: • An adult inhales 23 m 3 /day of air. • The body weight of an adult is 70 kg. • The inhalation absorption factor for DNC is 0.75. ------- B—V • The duration of exposure is 40 hours/week for a 30—year work period, or 10.2 percent of an average lifetime. Calculations: 0.047 mg (average DNC air concentration on-site) 23m 3 1 x x x 0.10 (human intake factor) day 70 kg x 0.75 (inhalation absorption factor) = 1.2 x mg/kg/day Inhalation of SOLVALL—Contantinated Air by Neighboring Residents Assumptions: • An adult inhales 23 m 3 /day of air. • Based on wind direction analysis, duration of exposure is 30 percent of time on an annual average basis. • The body weight of an adult is 70 kg. • The inhalation absorption factor for SOLVALL is 0.75. • The adult lives in the home throughout his lifetime. Calculations: 0.153 mg (average DNC air concentration at boundary) m 3 23m 3 1 x x 0.3 (human intake factor) x day 70kg x 0.75 (inhalation absorption factor) = 1.1 x 10-2 mg/kg/day ------- B—3 Inhalation of SOLVALL—Contazninated Air by Workers Assumptions: • An adult inhales 23 m 3 /day of air. • The body weight of an adult is 70 kg. • The inhalation absorption factor for SOLVALL is 0.75. • The duration of exposure to workers is 40 hours/week for a 30—year work period, equivalent to 10.2 percent of an average lifetime. Calculations: 0.7 mg (average SOLVALL air concentration on—site) m 3 23m 3 1 x x x 0.102 (human intake factor) day 70kg x 0.75 (inhalation absorption factor) = 1.8 x 1O2 mg/kg/day Ingestion of DNC-Contaminated Drinking Water at Eastern Property Boundary Assumptions: • An adult consumes 2 liters of water per day. • The body weight of an adult is 70 kg. • Absorption is 100 percent. • The water is consumed throughout the adultts lifetime. ------- B— 4 Calculations: 0.275 mg (predicted DNC concentration in ground water liter at eastern property boundary) 2 liters 1 x x (human intake factyr) day 70 kg x 1 (ingestion absorption factor) = 7.8 x i0 3 mg/kg/day Ingestion of DNC—Containinated Drinking Water at Public Well Field Calculations: 5.5 x i — mg (predicted DNC concentration in ground liter water at public well field) 2 liters 1 x x (human intake factor) day 70 kg x 1 (ingestion absorption factor) = 1.6 x mg/kg/day In estion of DNC—Contaminated Drinking Water at Point of Compliance Calculations: 3.32 x 10—1 mg (measured DNC concentration in ground liter water at point of compliance) 2 liters 1 x x (human intake factor) day 70 kg x 1 (ingestion absorption factor) = 9.5 x mg/kg/day ------- 8-5 Ingestion of SOLVALL-Contaminated Drinking Water at Eastern Property Boundary Assumptions: • An adult consumes 2 liters of water per day. • The body weight of an adult is 70 kg. • Absorption is 100 percent. • The water is consumed throughout the adult’s lifetime. Calculations: 7.7 x 10—1 mg ______— (predicted SOLVALL concentration in liter ground water at eastern property boundary) 2 liters 1 x x (human intake factor) day 70 kg x 1 (ingestion absorption factor) = 2.2 x 10-2 mg/kg/day Inyestion of SOLVALL-Contaminat Drinking Water at Public Well Field Calculations: 9.3 x 1O mg (SOtIVALL concentration in ground water liter at public well field) 2 liters 1 x x _ —__ (human intake factor) day 70 kg x 1 (ingestion absorption factor) = 2.7 x l0 mg/kg/day ------- B-6 Ingestion of SOLVALL—Contaminated Drinking Water at Point of Compliance Calculations: l.OxlO 0 mg (measured SOLVALL concentration in liter ground water at point of compliance) 2 liters 1 x x (human intake factor) day 70 kg x 1 (ingestion absorption factor) = 2.9 x 10-2 mg/kg/day ------- |