Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 Risk Assessment Exposure, laboratory, and epidemiological data provided earlier in this report are used in this chapter to make quantitative and qualitative (or comparative) assessments of risks from exposure to asbestifonm fibers. To place the discussion in context, the chapter begins with a brief general discussion of risk assessment and a few special considerations concerning asbestos and related fibrous materials. Various difficulties often limit the accuracy and precision with which risk to human health can be estimated. Nevertheless, when the data base is good, the risk est imate s can be suf f ic lent ly informal ive to aid policy judgments. Some of the factors that enhance the usefulness of the data include dose-response information based on several accurately known exposure levels; knowledge of physiologic and metabolic factors that affect exposure of body tissues; an understanding of the mechanism by which the substance results in toxicity; knowledge of the extent to which experimental systems mimic the human response; and an understanding of the properties of a complex and variable substance that account for its toxicity. Many of there issues apply in the assessment of risk from asbestiform fibers, which have varying physical and chemical properties. Some members of the class, the commonly used naturally occurring forms of asbestos, have been clearly shown to cause fibrosis of the lung and pleura as well as cancer of the lung, mesothelium, and possibly the gastrointestinal eract in humans. Some occupational data on other fibers are also available, and considerable numbers of experimental studies have been conducted. It is reasonable from a biological viewpoint to use data from occupational studies to derive estimates of risk from nonoccupational exposure. However, differences in route of exposure, type and characteristics of fiber, exposure levels, and time patterns must be considered. Moreover, because working populations are generally healthier than the public at large, the latter may contain a higher proportion of more susceptible individuals. THE PROCESS 0F RISK ASSESSMENT The principles guiding the assesament of health risks from environmental substances were recently reviewed by a committee of the 200
201 4 National Research Council ~1983~. These principles are summarized here to provide a framework for assessing the health risks from exposure to asbestiform fibers. The numerous terms uset to describe different aspects of risk assessment include 'hazard a~sesament," "hazard identification," "risk assessment," "qualitative risk a~sesament," "dose-response asee~sment," "comparative risk assesament," "quantitative risk assessment," and "risk characterization. " The use of these terms has not been standardized. Three concepts are generally incorporated into the risk assessment process. First is the identification of the kinds of harmful health effects, e.g., anemia, birth defects, or cancer, thee can result from sufficient exposure to a substance. Second is the do~e-response curve for a particular effect, i.e., the severity of damage and/or the percentage of people or animals likely to be at various exposure levels. Third is the number of people in a particular population, e.g., residents of the United States or workers in a particular industry, likely to be harmed under past, present, or projected levels and conditions of exposure. In this report, the committee has used "risk asses~mene" as a broad term encompassing all three of these concepts. "Hazard identification" refers to the first concept, "dose-response" curves or relationships are used in discussions of particular sets of data, and 'tquantiteeive rink assessment" refers to the estimates of risk to humans derived by mathematical extrapolations from there data. "Population risk estimates" describe the expected frequency or incidence of a harmful effect in a specific group of humans under defined conditions of exposure . The amount and complexity of informal ion needed increase as we progress from hazard identification to dose-response assessment to population risk estimation, although each step builds on the preceding one. Hazard identification characterizes the nature of toxic effects that a substance in capable of causing in laboratory animals or humans. Dose-response curves based on experimental or epidemiological observations define the frequency and sometimes the severity of these toxic effects at several levels of exposure. The dose-response information is used in quant itat ive risk estimation. Through mathematical modeling and application of known biological principles, attempts are often made to estimate risk for dose levels, exposure conditions, or species other than those for which do~e-response data have been obtained. For example, quantitative risk assessments often rely on dose-response data from studies of laboratory animals exposed to relatively high exposure levels in order to estimate the risk to humans exposed to lower levels. Assumptions and uncertainties involved in the application of quantitative risk assesament to cancer induction have been discussed extensively (Food
202 Safety Council, 1980; International Regulatory Liaison Group, 1979; Office of Technology Assessment, 1981~. Population risk estimates bring together quantitative risk estimates and data on exposure of a specific group of humane to identify their risk under actual or anticipated exposure conditions. The most relevant information for categorizing the hazard or the dose-response for humans is derived from studies of exposes humans. Unfortunately, evidence from this source is often unavailable or inconclusive at times when decisions about acceptable exposure must be made. Humans are exposed to so many dif ferent substances through food, medicines, air, water, household materials, and occupational environments that sorting out the causes of harmful effects on health is often difficult. Perhaps of most importance is the fact that evidence of human health hazards from substances introduced into our environment cannot be obtained directly from observations in humans until people have been harmed. For these reasons, evidence from laboratory animals or from other biological test systems is often used as an alternative or as a supplement to data on humans . A substantial body of evidence has demonstrated the utility of these experimental systems (Doull et al., 1980; National Research Council, 1977; Richmond et al., 1981~. A variety of mathematical models have been developed for using data at high doses, usually only available from studies in animals, to est imate risks for humans at low doses (Armitage, 1982; Cornfield et al., 1978; Crump et al., 1976; Fishbein, 1980; Food Safety Council, 1980; Krewski and Van Ryzin, 1981; Van Ryzin, 1980~. Because there are extensive data on the effects of asbestos and some other fibers in humans, the quantitative risk assessments in this chapter are based exclusively on data from epidemiological studies in humans, whereas the comparative risk assesement~ also take into consideration data from laboratory studies. Every scientific study or technique has some lower limit to its sensitivity. A sensitive method in analytical chemistry may be capable of detecting a few molecules of a particular chemical among a billion other kinds of molecules but incapable of detecting a few among a trillion. The sensitivity of an animal test for toxicity in limited by many factors, such as the number of animals that it is practical to study, the subtlety of the effect of interest, the occurrence of similar effects in animals not exposed to the material under test, and limitations on the Amounts of material that can be administered and on the methods used to administer them. Other difficulties limit the power of epidemiological studies. For example, it is often difficult to select appropriate control groups, estimate exposure, or detect health effects from the exposures of concern, especially if the exposures are much lower than those that occur among occupat tonal groups .
203 . 1 1 Several kinds of information are use fut for estimating risks at low exposure leirele on the basis of observations at higher exposures. These inc. lude the shape of the dose-re sponge curve in the range of exposure studied, knowledge of the mechanism by which the type of toxic effect occurs , and informal ion on dose-related changes in the uptake, distribution, chemical or physical modification, and exereeion of the substance, i.e., phanmacokinetica. - Substances vary "Arkedly both in the quantity required to produce a toxic effect and in the rapidity with which the incidence of toxic effects decreases with decreasing dose, i.e., the shape of the dose-response curve. In an experiment covering a sufficiently wide range of exposure levels, it is possible to find some levels that are toxic and some lower leve Is at which no toxic ity is observed. me highest dose at which no toxicity is seen is often called the "no-observed-effect level, " or NOEL (Klaassen and Doull, 1980) . However, any experiment will have some limit in its sensitivity to Small effects, ant the true no~effect-level, if any, may be below the NOEL in a part ice tar experiment . The fundamental assumption underlying the NOEL safety factor approach is that Some minimal level of a toxic substance is required to cause damage and that the substance is not toxic below that level. The NOEL type of experiment is used to f ind that leve 1. The maximum dose at which no toxicity would occur is called the "threshold" for that substance. However, several mathematical models for quantitative estimation of cancer rick assume that there is no threshold; risk diminishes with decreasing dose, but some risk in assumed to remain as long as there is any exposure. The determination of which of these two assumptions is correct will probably depend on the nature of the toxic effect. Thus, understanding the mechanism of toxicity can provide guidance in seeting acceptable exposure levels. For a substance that exerts its toxic effect by inactivating an enzyme present in abundance in each cell, it is reasonable to assu~ that a threshold would exist. Inactivation of a few molecules of the enzyme is unlikely to damage the cell. On the other hand, a chemical that is mutagenic or carcinogenic because it damages some critical site on a DNA molecule that starts the carcinogenic process can reasonably be aced not to have a threshold. The likelihood that a critical site would be damaged would decrease with decreasing dose, but the possibility that this damage could occur rema ins at any exposure above zero . For many effects, the severity of the toxic effect, as well as the probability that it will occur, also decreases with dose. For example, a dose that damages a high proportion of cells in the liver may be lethal; one that damages a moderate number may cause severe illness but not death; a Small dose that causes damage to a few cells may not lead
204 to any clinical symptoms. The error in assuming a threshold if none truisr existed would generally not be expected to lead to serious cases of disease in this situat ion. By contrast, the severity of cancer and of mutat ions is not re lated to the dose of the substance causing them. Low tose exposure to x-rays or cigarette smoke causes fewer cancers than does high dose exposure, but the resulting cancers are just as lethal. Emus, although there may be some substances that show a threshold for cancer induction (Hoer et al., 1983), an error in assuming a threshold when none really exists would severe ly harm those persons who got the disease despite a low exposure . Accurate documentation of exposure is important for determining the dose-response curves for toxicity in animals or humans and also for estimating population risks. Errors in the estimation of exposure will lead to errors in He f ining the dose-re spouse curve and in making quantitative risk estimates for individuals or specific populations. The amount of a toxic substance or its active metabolize that reaches the body site that is susceptible to its effect is the exposure that accounts for toxicity, but such measures are almost never available (Hoer et al., 1983~. Other measurements, such as amounts in the blood, amounts entering the body, or concentrations in the air or water of a community, are often use ful surrogates , but as noted earlier in this report, they are also often unavailable. The sensitivity of the exposed population is another consideration in the risk estimation process. Some individuals may be more sensitive than others to specific environmental insults because of nutritional deficiencies, genetic predisposition, and for children, small body size, developmental immaturity, and increased metabolic and respiratory rates (Calabrese, 197B, 1980) . With their rapid metabolic rate, children consume proportionately more food and inhale greater volumes of air than an adult for a given body weight. Thus, they would also consume or inhale proportionately more of any contaminants that are present (8abich and Davis, 1981~. Human infants do not have mature hepatic detoxification systems until they reach 2 to 3 months of age (Pelkonen et at., 1973; Rane and Ackerman, 1972~. Serum in~munoglobulin does not attain adult levels until children are 10 to 12 years old (Calabrese, 1978~. Studies in animals have also demonstrated a greater sensitivity among the young after exposure to chemicals by a variety of routes (Goldenthal, 19711. Children's lunge may also be especially sensitive to environmental pollutants. Tager et al. (1983) have observed measurable differences In lung function between children of smoking mothers and children whose mothers did not smoke.
205 i Population risk estimation is based on all the preceding steps. First, the exposure of the study population must be known. Heterogeneity of the population with respect to level of exposure or sensitivity to the toxic material should also be considered in the calculations. Exposure, dose-response curares, distribut ion of sensitivity factors, and the size of the population are then used to estimate the number of people likely to suffer toxic effects from the substance of interest. If the material causes more than one type of toxic effect, each effect requires separate calculations. Ideally, calculation of risk is an objective, scientific activity devoid of policy judgmenta. The latter are made separately when deciding the acceptable level of exposure. However, policy decisions can seldom be divorced completely from the process of risk asse~ement. The reason for this lies in the uncertainty of many of the scientific judgments required. For example, if one experimental species is more susceptible to the toxicity of a material than another and data on humans are unavailable, which species should be used for estimating human risk? Which mathematical model should be applied to the data? These and many other questions of judgment were discussed in the recent National Research Council (1983) report. In the following sections, the committee has used epidemiological data, mostly from occupational settings, to develop a quantitative model of the relationship between fiber dose and carcinogenic response for a generalized "asbestos" exposure resulting in em er lung cancer or mesothelioma. That dose-response relationship is then applied to a hypothetical, but reasonable, exposure level to show potential population risk levels in populations of arbitrary size. In the final section, the committee assesses ricks for other types of fibers and, in same cases, for other tiaeases by qualitative comparisons with the base case of a generalized asbestos exposure. QUANTITATIVE RISK ASSESSMENT In the previous chapters, the commit tee extent ive ly reviewed information on the health effects of asbestos and other asbestiform fibers. In preparing this section, it also reviewed several risk assesomento for asbestos in the open literature and in government documents. On the basis of its evaluation of the quality and coverage of the information and the as~esoment techniques, the committee decided that a quantitative assessment of the risk- for mesothelioma and lung cancer from nonoccupational exposures to asbestos would be meaningful. It also concluded that the information base was insufficient for useful quantitative assesamenta for other fiber types and tiaeasea, but that in some cases a qualitative, comparative assesoment was feasible and useful. me se decisions do not mean that the asbestos assesoment is without major uncertainties nor does it mean that the comparative assessments are of poor quality. In both cases, the ob jective is to
206 present information useful for evaluating the health risks of asbestiform fibers in nonoccupational settings. First, an overview of mathematical models for carcinogenic risk assessment in presented to provide a context for the assessments for lung cancer and mesothelioma, which are of principal interest. Next, there is a review of several assessments for asbestos that were based on such models. Finally, these assessments and the committee.e own analyses are applied to the information presented in earlier chapters to produce quantitative risk estimates for nonoccupational exposures to asbestos in ambient air. Mathematical Model for Carcinogenic Risk Estimate As explained earlier, it is not necessary to use data on asbestos exposure from animal experiments to estimate risks for humans, but it is necessary to extrapolate from the health effects observed at high occupational levels of exposure to much lower nonoccupational exposures. Occupational epidemiology makes it possible to describe the probability of dying from a particular type of cancer as a function of age at first exposure, level and duration of exposure, and current age. Mathematical extrapolation models based on the multistage theory of carcinogene~is make it possible to estimate the probability of dying from that type of cancer for different ages at first exposure, different (lower) exposure levels, and different (often longer) duration of exposure, also as a function of current age. By considering the cumulative probability throughout a lifetime, the "lifetime risk" of cancer mortality can be computed. At any age, an individual faces some probability of reaching an end point that is related to cancer in the next year, for example, dying of lung cancer. Suppose that at a given age, a, the probability is given by pta,d), where d is the dose of the carcinogen--in this case, asbestos. When d = 0, p~a,O) is the probability of the end point for unexposed people. If t is some age of interest, then the cumulative probability P(t,d) of reaching the end point before that age is given by the sum of the annual probabilities up to that age: P(t,d) ~ the sum of plead) over all ages, a, At. (1) Reaching the end point by time t is analogous to the "failure time" for a generalized system that is no longer effective after time t. General mathematical analysis can be used to show thee the probability of failure as a function of time can be written as follows: P(t,d) = 1 - e -I(t,d) (2) where I(t,d) represents the cumulative incidence function (or cumulative hazard function) of occurrence of the observable failure prior to time t.
207 Armitage and Doll (1961), Peto et al. (1982), Kalbfieisch and Prentice (1980), Hartley and Sielken (1977), Hartley et al. (1981), and RalLfieisch et al. ~1983) have applied this mode} to carcinogenesis. If the cumulative incidence Itt,d) is small, then equation (2) may be simplified to P(t,d) ~ I(t,d), where - means approximate ly. (3) In carcinogenic risk assessment, attention is usually focussed on the cumulative incidence function I(t,t) rather than on the probability function P(t,d). The Armitage-Doll (1961) multistage theory of carcinogenesis suggests that I(t,d) can be written as a product of two terma--g~d) , depending only on dose, and hits, depending only on time. That is, I(t,d) ~ gods hits. 1 (4) If there are k dose-dependent stages in the process of carcinogenesis and the rate of transformation from one stage to the next in assumed to be a linear function of dose, the function gods would be a polynomial of degree k in the done. The function htt) depends only on time. This mode! and its generalization and justification have been discussed by Grump et al. (1976), Hartley et al. (1981), and Kalbfleisch et al. 1983~. To determine the values of the constants in the polynomial gods and the functional form for htt), the cumulative incidence function must be fitted to daea--preferably to data based on observations in human populations. The multistage model described above has been fitted successfully to many sets of cancer data, including data on asbestos, and appears at present to be a generally adequate model for assessing cancer risk. Fitting equation (4) to data involves estimating the constants in the mode! for some suitably determined function htt). This mode! has been applied to both mesothelioma and lung cancer data on asbestos~exposed workers. The form of htt) and ache values of the constants from those studies will be discusses in the next section. The function g~d)--and thus the cumulative excess incidence function I(t,d)--can be approximated as a linear function of dose in the low-dose range that equals O when d ~ 0. This relationship can be used for extrapolating from high to low doses and has ache following form: I(t,d) ~ cdh(t). (5) Lois form assumes that there is at least one do~e-depentent stage of cancer development. The argument for a linear (with respect to dose) approximation for low-dose exposures has been justifies on the basis that the exposure dose d is added to a background leve ~ (Hoe I, 1980; Peto, 1978~. This assumption may not always be justified in application
t 208 (see Cornfield et al. ~ 1978 and Van Ryzin, 1981), but it should lead to an appropriate upper bound for the co~ittee's risk assesements for asbestos. Furthermore, and more importantly, ruling out a linear dose term for asbestos exposure does not seem justifies by the data now available (Nicholson, 1983; Peto, 1982; Schneiderman et al., 1981~. Thus, the mode ~ adopted for risk assessment in the next three see t ions of this chapter is based on the cancer mortality incidence calculated by equse ion (5 ~ . PUBLISHED RISK ASSESSMENTS This section reviews some published risk assessments for lung cancer ant mesotheliorea. me se assessments helped the committee select a functional form for hi t) for the two diseases and to establish the value of the constant c in equation (5~. Lung Cancer Rick from Nonoccupat tonal Environmental Exposures The following summary of risk a~sesamento for lung cancer from asbestos exposures is based on data on exposure of worker populations. These data suggest that the function I(t,&) in equation (5) tee corset I(t,d) ~ c*TodIof t), (6) where To is the durat ion of exposure to asbestos at dose d, Ion t ~ is the cumulative mortality incidence for lung cancer up to age t for those who have not been exposed to asbestos, and c* is a constant that depends on the cohort under study, but not on dose or age. As used in equation (6) and in the remainder of this section, d in the concentration of fibers in the workplace air, usually measured in fibero/cm3. Although d is referred to as dose, some authors would call it done rate and would refer to the product Ted as (cumulative) dose. Equat ion (6), derived by Peto ( 1982), is consistent with his earlier studies of chrysotile workers (Peso, 1978~. This equation is also supported by four studies reviewed by Nicholson (1983), who noted that the relative rick of lung cancer deaths for asbestos workers compared to a similar population was linearly related to the accumulated dose years, i.e., fibers/cm3 x years, or (fibers/cm3)yr. In equation (6), the underlying incidence rate Iott) is consider- ably different for smokers and nonsmokers of each sex. Therefore, the risks for each of these groups must be assessed separately. Another consequence of equation (6) is that the relative risk of lung cancer due to asbestos exposure does not depend on age at first exposure. Thus, lifelong risk of lung cancer resulting from exposure to asbestos can be calculated quite simply by using equation (6~. As an example, consider the following calculation given by Peto (1982~. -
209 Consider the effect of 10 years of exposure at 1 fiber/cm3. If we assume thee the relative risk for lung cancer among insulation workers increased approximately fourfold tHa~ond et al. ( 1979) reported 4.2 for nonsmokers ant 3.9 for smoke ret ~ and that this risk is based on a cumulative dose of 600 fibers/cm3 (20 years at 30 fibers/cm3), then 10 years of exposure to 1 fiber/cm3 will increase the relative risk by 4.0 x 10/600 ~ 0.067. Since approximately 15% of lifelong smokers die of lung cancer, this mortality rate will increase to 0.15 x 1.067 x 100, or 16%. mus, the difference (1X) is the excess due to asbestos as predicted by the equation. Since only 0. 5% of nonsmokers die of lung cancer, this would become 0.533% (0.005 x 1.067 x 100) for an added risk of 0.033: due to asbestos exposure. Mesothelioma Risk from Nonoccupational Environmental Exposures The committee reviewed two estimations of mesothelioma risk, one by Peto and his colleagues (Peso, 1982; Peto et al., 1982) and the other by Nicholson (1983~. These analyses and their consequences are summarized in this section. Using the data of Selikoff et al. ( 1979) on mortality among 17, 800 members of the Internat tonal Assoc fat ion of Heat and Frost Insulators and Asbestos Workers, Peto et al . ( 1982 ) showed that the mortality rate from mesothelioma in these workers was dependent on the time since first exposure, but did not depend on the age at first exposure. From this finding, and the application of the multistage theory of carcinogenesis through equation (5), the cumulative incidence function becomes: I(t,d) = edit - talk, (7) where t - to represents time since first exposure at age to. For any group of workers exposed at the same dose leve 1 d, the produc t cd = b is a constant depending on the type of asbestos exposure . Equat ion ~ 7 suggests Chat the risk for mesothelioma is primarily dependent on the time since first exposure (t - to). This same phenomenon was noted by Schneidennan et al. (1981) and Nicholson (1983~. Fitting equation (7) with b = cd to the data of Selikoff et al. ( 1979) for men up to age 80 by the method of maximum likelihood estimation resulted in an estimate of k = 3.2 with a standard error of + 0.36 and b = 4.37 x 10-8. Using this calculation, Peto et al. (1982) estimated the lifelong mesothelioma risk for this worker group to be 15:, It, and 3% for age at first exposures of 20, 30, and 40 years, respective ly. These figures have been adjusted for other competing causes of teeth. Using equation (7) witch k a 3.2, Peto and colleagues determined that b x 108 ranges in value from 2.94 to 5.15 for four other sets of data (see Table 7-~. Using k ~ 3. 5, PeCo (1982) computed a lifetime mesoehelioma rate of 1 in 100,000 children exposed from age 12 to age 18
210 TARlE 7-~. Mesothelioma Death Rates in Various Studies and Predictions of Riska Study Population Relative Risk and Reference (b s 108) Corresponding Lifetime Risk (~)b byAgeat First Exposure (yes ) 20 30 40 North American insulation 4.37 15 7 3 workers (mixed exposure) Selikoff et al., 1979 Factory workers (mixed 4.95 17 ~3 exposure) Newhouse and Berry, 1976 Chrysotile textile 2.94 factory workers Pe to, l980b 10 5 2 Australian crocidolite 5.15 17 ~3 miners Hobbs et al., 1980 U.S. amosite factory 4.91 17 ~3 workers Seid - n _ al., 1979 aAdapted from Peto et al. (1982~. The death rate at time t - to since first exposure at age to is proportional to b, obtained by fitting equation (7) with k ~ 3.2. bThe calculation of "lifetime risk," i.e., the percentage of similarly exposed men who would die of mesothelioma before age 80, is based on an actuarial calculation using 1977 U.S. rates for white males for all causes of death other than mesothelioma inflated by a factor of 1.26, the observed relative risk among insulation workers (Selikoff et al., 1979~. (i.e., 6 years of school age), assuming the fiber level was 0.003 fiber/cm3 (~/l, 000 of the exposure of the insulation workers). A second risk assessment was done by Nicholson (1983), who criticized the Peto et al. (1982) analysis for fitting equation (7) to only those men who died of mesothelioma up to age 80. By including all insulation workers, he estimated k to be 5.0. 1
:' 211 QUANTITATIVE RISK ASSESSMENT FOR NONOCCUPATIONAL ENVIRONMENTAL EXPOSURES As a starting point for assessing the risk from nonoccupational environmental exposure to asbestiform fibers, the committee adopted equation (6) as representing the cumulative mortality up to age t, which is appropriate for lung cancer induced by a continuous exposure of To years at dose level d in fibers/cm3. This model implies that any given total dose before time t would have the same effect on the relative risk at time t, regardless of the time at which exposure started or its duration. The model thus ignores a minimum latency period, which might cause the model to overestimate effects, but also ignores the difference between exposures at earlier and later ages, which might cause the model to underestimate effects. Equation (7) was assumed to be a reasonable representation of the cumulative mortality from mesothelioma up to age t for continuous exposure to asbestos at dose level d in fibers/cm3 from age to until age t. In this case, latency is implicitly included in the dependence on (t-to), because k is greater than 1, but no minimum latency is assumed. mese assumptions are supported by the work of Peto (1982), Peto et al. (19B2), Nicholson (1983), and Schneiderman et al. (1981), who extensively reviewed the basis for these assumptions by examining the models and their consistency for several observed worker cohorts exposed to ambient concentrations of asbestos fibers. These authors have suggested that asbestos acts as a late-stage carcinogen in producing lung cancer but acts at earlier stages in the development of mesothelioma. Using these models, the committee developed lifetime estimates of risk for lung cancer and mesothelioma mortality from continuous nonoccupational exposures to 0.0004 fibers/cm3 and for 0.002 fibere/cm3 . For lung cancer, the committee assessed the risk for four exposure subgroups: male smokers, female smokers, male nonsmokers, and female nonsmokers. For mesothelioma, only one calculation was made, since equation (7) and the supporting data in the papers cited above suggest that mesothelioma mortality does not depend on sex or smoking history, but does depend ~ trongly on age at fire t exposure. Life time Risk Estimates for Lung Cancer and Mesothelioma Table 7-2 summarizes lifetime risk estimates for lung cancer and mesothelioma for nonoccupational environmental exposures to 0.0004 fibers/cm3 (a median level) and 0.002 fibers/cm3 (a high level). It is assumed this exposure is continuous from birth through a lifetime of 73 years, an approximate average lifetime in the United States. mus, in equations (6) and (7), t ~ 73 years and d a 0.0004 or 0.002. In equation (6), To ~ 73 and in equation (7), to ~ 0 to account for continuous exposure. Because equations (6) ant (7) are linear in the dose unit d, one can immediately obtain from Table 7-2 lifetime risks at other continuous (from birth) environmental exposures by multiplying by the appropriate dose factor. For example, lifetime risk estimates at 0.02 fibere/cm3 are 10 times higher than the estimates at 0.002 fibers/cm3.
212 TABLE 7-2. Eatimated Individual Lifetime Riske from a Continuous Exposure to Asbestos at 0.0004 Fibers/cm3 (a Median Dose) or 0.002 Fiber s/cm3 (a High Dose ja ~L L_~ ~ tOb Exposure Median Exposure Iligh Exposure Disease Group (0.0004 fibers/cm3) (0.002 fibers/cm3 Lung cancers Male smoker 64 (0 to 290)C 320 (0 to 1,500) Lung cancer Female smoker 23 (0 to 110) 120 (0 to 530) Lung cancer Male nonsmoker 6 (0 to 22) 29 (0 to 130) Lung cancer Female nonsmoker 3 (0 to 13) 15 (0 to 66) Mesothelioma All 9 (0 to 350) 46 (0 to 1,700) . aLifetime assumed to be 73 years; exposure occurs from berth. Lung cancer ricks are calculated with c* ~ 1.02 or an excess risk of 2: per (fiber/cm3)yr, estimated from nine studies with varied results. Mesothelioma risks are calculated with c ~ 2.53 x 10-8 and k ~ 3.2, estimated from five studies with varied results. See also explanations in text. bSex differences for lung cancer risk are due to differences in lung cancer background rates associated with smoking patterns, occupational exposures, and other factors. CRange of estimates. The lower limit of O is always possible if linear extrapolation overestimates risk. See also text below. The estimates in Table 7-2 were based on the following five considerations: · Exposure levels. A mix of indoor ant outdoor measured exposure levels was used to select the median value of 0.0004 fiber6/cm3 and the high value of 0.002 fibers/cm3 as the reference levels. · Use of the linear model. The models used by the committee all assume low-lose linearity and, as such, produce higher estimates of risk at low doses than would be obtained with other models. However, because the occupational data do not rule out low~tose linearity, the committee believes that these estimates do not unduly overstate the risks. · Countless conversion. We conversion of ambient fiber mass measurements to an equivalent number of fibers was based on measurements
213 of mass and awmbere of fibers in the workplace. The committee realized, however, that the number of fibers in ambient air would be much greater because these fibers tend to be smaller than those in the workplace (see Chapter 4~. Depending on the toxicity of small fibers, the risks could be greater or less than those calculated in this chapter. If the presence of long fibers is necessary for a toxic response, risks would be lower. ~ Model dependence. The results of the mesothelioma model depend very heavily on the value of k. This accounts for the large range of estimates for mesothelioma. It is assumed that this dependence on k among workers holds for the entire population throughout a lifetime. If the dependence is not as strong (i.e., a lower k value), the lower end of the range would apply. If this dependence is as strong (i.e., a higher k value), the upper bound may be more appropriate. · Childhood exposure. The models used for extrapolation for both lung cancer and mesothelioma are based on the assumption that a unit dose of exposure (measured as fibers/cm3 > 5 Am long) in early life is equivalent in its intrinsic care inogenic potential to a unit dose in later life. If children are more biologically sensitive than the worker group, the risk per unit dose would be increased. Results from studies of exposure to other materials indicate that children are often more sensi- tive than adults to a given dose, even when expressed as dose/body weight. The risk estimates and ranges shown in Table 7-2 are those the committee considers most reasonable. Because of the uncertain value of k and the sensitivity of equation (7) to its value, the range of estimates is much larger for mesothelioma than for lung cancer. Two conclusione can be drawn from the estimates in Table 7-2: o For nonsmokers, the lifetime risk for mesothelioma from non- occupational environmental exposure to asbestos is higher than for lung cancer. For smokers, however, the risks of lung cancer are substantially higher than for mesothelioma, because of the multiplicative interaction of Smoking and asbestos exposures. · Individual lifetime risk estimates for lung cancer from nonoccupational environmental exposures to 0.0004 fibers/cm3 are much lower than the risks observed for smoking. me basis for the calculations in Table 7-2 is discussed in detail in the following two subsections. Calculation of ~ . Calcu- lati ~ involves the notion of relative risk up to time t, designated here as RR. From equation (6), the RR for lung cancer by age t can be shown as follows:
t I(t,d) Io(t) 214 cumulative lung cancer mortality by age t at dose d baseline cumulative lung cancer mortality by age t = c*(Tod), where (Ted) ~ total dose-years for the exposed group and c* is a constant that depends on the cohort. Ad) For a given study showing an increased relative risk for lung cancer, c* a (1 + P/100), where P is the percentage increase in lung cancer risk per unit dose [X per (fibere/cm3)yr]. Schneiderman et al. (1981) presented the values of P for nine different worker cohorts. The results are seized in Table 7-3. (9) Values for P in Table 7-3 range from 0.06 (Study 8) to 9.1 (Study 1~. me higher value establisher the upper end of the range given in Table 7-2. m e zero value for the lower end of the range indicates that the low-dose linear approximation in equation (5) may overstate risk. The median value for P in the studies shown in Table 7-3 is p a 1.1 (Study 7~. mid value, rounded upward to 2, was uset in obtaining the estimates for lifetime lung cancer risk in Table 7-2. To calculate these estimates, it was necessary to know only the baseline absolute risks for the appropriate subpopulations. The baseline cumulative incidence rates of lung cancer for the four subgroups in Table 7-2 have been estimated by Schneidenman_ al. (1981) as follows: male smokers = 0.11; female Smokers a 0.04 ; male nonsmokers ~ 0.01; and female nonsmokers ~ 0.005. Thus, using 2: as a value for P. the lifetime risk of lung cancer for a male smoker is (0.11)(1 + P/100) ~ (0.11)(1 + 0.02) a 0.1122. (10) The increased lifetime risk attributable to asbestos exposure at ~ fiber/cm3 for ~ year is 0.0022, i.e., 0.1122 - 0.1100. At the Ambient exposure of 0.0004 fibers/cm3 assumed in Table 7-2 and for a 73-year lifetime exposure, the increased lifetime risk of lung cancer in 6.42 x 10-5, i.e., 0.0022 x 0!0004 x 73. Rounding to two significant figures gives the estimate in Table 7-2 for male smokers. me other calculations in that table were derived in a ~ imi tar fashion. When describing the use of the percentages given in Table 7-3, Schneiderman et al . ( 1981 ) commented that the low percentage increases in risk in Studies 3, 6, 8, and 9 probably resulted from several factors. In Study 3, for example, the subjects were retirees older than 65. 1
215 TABLE 7-3. Estimated Increase in Lung Cancer Risk per Unit of Exposure to Asbe s tosa Study Occupation of No. Worker Cohort Asbestos Type 4 Asbestos produc t manu facturing i 6 5 Percent Increase in Lung Cance r Ri sk pe r ~ f ibers/cm3 Or Re ference 1 Insulation Amonite 9.1 Seidman et al., manufacturing 1979 2 Asbestos Crocidolite, 1.3 males Newhouse and product manu- chrysotile, 8.4 females Berry, 1979 factoring and amosite 3 Asbestos manufac Luring Text i le produc t ion Text i le product ion 7 Insulation Chrysotile 1.7 manufac- and amosite turing Amosite and chrysotile; some crocidolite 0.3 Chrysotile; 1.1 some amosite and crocidolite Chrysotile 5.3 Henderson and Enterline, 1979 Nicho 1 son _ al., 1979 cement et al., 1982 Chrysotile 0.07 early Peto, 1980 employeesb 0.8 later employeesb Selikoff et al., 1979 8 Mining Chrysotile 0.06 McDonald and and milling Liddell, 1979 q ~ ning Chrysotile 0.15 Nicholson and milling et al., 1979 aAdapted from Table 4 in Schneiderman et al., 1981. bEarly employees began work before or during 1950. Later employees began work af ter 1950.
216 Schneidenman et al. stated that the investigators may thus have missed asbestos-related~~eatho occurring at earlier ages. In Study 6, the disease rates for workers employed earlier were lower than those employed later who were followed for aborter periods. The discrepancy has diminished as more data have accumulated. The subjects in Studies and 9 were mining and milling workers whose exposure patterns were quite different from environmental ambient air exposures. There in also some evidence that many lung cancer cases were missed in Studies 8 and 9 because of competing causes of death at earlier ages. Thus, Schneidenman et al. (1981) concluded that the range from I.1 (Study 4) to 9.1 (Study 1) is the most representative of true values. The value of P ~ 2 used in the calculations in Table 7-2 falls near the bottom of this range, but is within a factor of 5 of the top of the range. If we use P = 5, which in the middle of the range, trebling cancer risk estimates in Table 7-2 would be multiplied by a factor of 2. 5. Calculation of Mesothel~oma Risk Estimates. To calculate the lifetime risk with equation (7), the numbers c and k must be deter- mined. Then the lifetime risk L at d a 0.0004 fibers/cm3, assuming t = 73 and to = 0 (continuous exposure from birth to age 73), is L = c(0.0004~73)k. (11) To apply this equation, c and k must be estimated from epidemiological studies of occupat tonal exposures to asbestos . Each study must be stratified by duration of exposure (t-to) to estimate there parameters. Most of the following analysis is similar to that of Peto _ al. (1982~. First, let us consider the choice of k. As noted earlier, when Peto et al. (1982) fitted equation (1) to the data of Selikoff et al. (1979), they obtained the equation I(t,d) ~ bit - to)3 2, with b = 4.37 and k = 3.2 + 0.36 (standard error). In equation (11), therefore, we initially use k = 3. 2. Modifications using different values for k will give the range of est imates for d = 0.0004 fibera/cm3 in Table 7-2. For d = 0.002 fibers/cm3, we replace 0. 0004 with 0.002 in equation ( 11) . With k ~ 3.2, Peto et al. ( 1982) also fitted four other data sets to obtain four values of b in the equation I(t,d) ~ bit - to)3 2. me value of b is specific to each worker cohort and depends on three numbers : d (the average fiber/cm3 exposure), ~ (the average length of exposure), and e - to (the average time since first exposure). These values are given in Table 7-4. In addition, Table 7-4 contains the estimates of c that are appropriate for equation (7), based on the corresponding estimate of b given by Peto et al. (1982~. When exposure is not continuous from time of first exposure (to) to the age of observation (t) for these studies, the relationship between b and c changes from c ~ bid to 4.56 bid 1 _ (12)
l 217 TABLE 7-4. Estimated C088ta~t8 for Equations (~) and (12) for Five Studies Study b ~ 108 Ma Ma t - to a c s 108 Selikoff et al., 4.37 15 15 24 1.39 1979 Newhouse and 4.95 12.5 6 31.5 3.67 Berry, 1976 Peto, 1980a,b 2.94 16.5 14 22.5 0.85 Hobbs et al., 5.15 NAb NA NA NA 1980 Seidman et al., 4.91 35 1 35 7.22 1979 aEstimated from data given in Tables 4 and 10 of Schneiderman et al. (1981), using estimated median values. The product dQ from C3i~B 3 and 4 above is the estimated cumulative exposure in (fiber/cm )yr of their Table 10. bNA ~ not available. The factor 4.56 adjusts from occupational exposures at about 1,920 hours per year to environmental exposures at 8,760 hours per year. Appendix G provides the mathematical basis for equation (12~. Table 7-4 gives the values of the constants for each study in which Peto et al. (1982) estimated b. To obtain the estimates for mesothelioma at the dose of 0.0004 fibera/c.3 in Table 7-2, equation (~) is used with values for c from Table 7-4 and k - 3.2. In Table 7-2 the lifetime risk for mesothelioma at d ~ 0.0004 fibere/cm3 is 9 per million. This is calculated from equation (~) with c - 2.53 s 10~8, the median of the range of the c values in Table 7-4, and k - 3.2. The highest value of the binge in Table 7-2 at d ~ 0.0004 uses equation (~) with c ~ 7.22 s 10~ , the upper value of c in Table 7-4, and k ~ 3.S, obtained from 3.2 + 1.65 s 0.36. The selection of 3.S as the value for k is based on an approximate upper 95Z confidence limit for the estimate of k. The lower limit is taken as 0, which is always a possible lower limit, especially if the low-dose ilnear assumption in equation (5) overestimates the individual lifetime risk.
~ - 218 Peto ( 1982) recommended using a k value of 3.5 for risk assessment purposes. As an example, he estimated that the rick of mesothelioma for children exposed for a 6-year period (ages 12 deco 18) arc 0.003 fibers/cm3 would be one in 100, 000. Nicholson reviewed additional data, including data on older workers up to age 80, and determined Chat a k value would be 5. Scl~neiderman et al . ( 1981 ) used k a 3. O. For this study, the committee used a value of 3. 2. Although neither exist ing data nor biological theory can provide very much guidance on the value of k, its value is very important in projecting the lifetime risks of mesothelioma from asbestos exposures. Table 7-5 shows how lifetime risk varies from the value of 9 per million for several values of k. Also shown are risk estimates for other values of c. The reader can easily calculate the results for other values of exposure. Other authors have also estimated the risks of mesothel iomas. Enterline (1983) derived a lifetime risk of 100 per million by using current reported rates of mesothelioma, an assumption about the relative contributions of nonoccupational and occupational asbestos exposures, and other factors. This estimate clearly relater to past exposure to varying levels of asbestos. Schneiderman et al. (1981) estimated lifetime risks for mesothelioma to be between 800 and 5,000 per million for a cumulative exposure of 1 (fiber/cm3)yr. These estimates correspond to lifetime risks of 23 to 150 per million for 0.0004 fibers/cm3 for 73 years. As mentioned above, these investigators effectively assumed k a 3, but their equivalent c was higher than that used for the corresponding estimates in Tables 7-2 and 7-5. TABLE 7-5. Sensitivity of Estimates for Lifetime Riskea of Mesothe 1 ions to Values of k and c Lifetiae Risk Estimates x 106, Using k Values from Various Studies - Peto et reto et This Study Schneider This Study al., 1982 This Study al., 1982 Nicholson, (low) et al., 1981 (middle) 7iiiddle) (high) (high) 1983 k c 2.6 3.0 3.2 3.5 3.8 4.0 5.0 0.85 0.2 x 10-8 2.53 0.7 s 10-8 7.22 x 10-8 t.3 3 11 41 97 7,000 4 9 34 120 290 21,000 2 11 26 96 3S0 820 60,000 .411 estimates are derived from equation (11), L - c(0.0004)(73)k, where L - lifetime rifle at a continuous exposure to 0.0004 fibere/ce3 for a lifetime of 73 ~care. Note: This table demonstrates that the rislc estimates are extremely acnsitive to changce in the value of k.
219 i The Use of 0.0004 Fibers/cm3 and 0.002 Fibers/cm3 as the Median and High Nonoccupational Environmental Exposure Levels. The lifetime risk estimates given in Table 7-2 are based on an assumed continuous environmental ambient exposure equivalent to either 0.0004 or 0.002 fibers longer than 5 Am per cm3 of air breathed. The committee believes that 0.0004 fibers/cm3 is a reasonable assumption for a median population exposure level and that 0.002 fibers/cm3 is a reasonable high exposure level (considering only exposures from breathing ambient air continuously). mese assumptions are discussed below. The effects of noncontinuous high exposures are discussed later in this chapter. Table 7~6 summarizes some environmental asbestos sampling data provided by Nicholson (1983~. To convert from mass measurements (r~g/m3) of airborne exposures to fiber counts (fibers/cm3), the committee used the conversion factor of 30 pg/m3 for 1 fiber/cm3. (See Chapter 4 of this report, Schneiderman et al., 1981, and Consumer Product Safety Commission, 1983 for further explanation.) The dose-response data used in the committee's risk estimate were taken from measurements of exposures in the workplace, where the fibers tend to be longer than those in ambient environments not close to major sources of asbestos. As discussed in Chapter 4, there would typically be approximately 2,000 fibers per nanogram in workplace air; in remote areas, however, there would be approximately 70,000 ambient fibers in a nanogram. To convert mans in the workplace to ambient air, the committee used the number of fibers longer than 5 Am that would be found in the workplace when the workplace mass equaled the remote ambient fiber mass. The dose estimate in numbers of fibers would be approximately 35 times greater (70,000/2,000) if the actual sizes of fibers in ambient air were considered. If we assume that all fibers are equally potent, then the risk estimates would be correspondingly higher. On the other hand, fiber size apparently affects fiber potency, but the appropriate adjustment factors for fiber size are not known. Table 7-6 indicates that median concentrations in outdoor air have ranged from 0.00002 to 0.00075 fibers/cm3 in several studies (sample Bets 1 to 8); their median is approximately 0.00007 fiber6/cm3. The observed median inside rooms without asbestos is 0.00054 (sample set 93. In rooms with asbestos surfaces, the median is 0.0006 fiber6/cm3 (range of medians for sample sets 10 through 14, 0.00006 to 0.00405 fibera/cm3~. If these three medians are weighted by assuming persons spent approximately one-fourth of their time outdoors, five-eighths of their time indoors in uncontaminated rooms, and one-eighth of their time in asbestos-contamirmted rooms, a reasonable estimate for a median population exposure is 0.0004 fibers/cm3. The committee also used 0.002 fibera/cm3 for a high value of continuous exposure in its calculations for Table 7-2. This value was obtained by using the median of the 90th percentiles in Table 7~6 for each exposure subcategory. For outdoor air, the median is 0.0003
220 lb. of Sample Set · Samples 1. Paris air 161 2. Paris (outdoor control) 3. Outdoor control susplce, for U. S. school ~ ' . Air of 48 U. S. cities 5. Air of V.S. cities 6. Air of five U. S. c it is ~ (outdoor control supple) 7. New York City air TILE 7~6. Spry of Baviro~ente1 "bestoe ~pocurc Samples Heasured Concentra- Equi~raleat Concentre tion (n.,/~3) lion (fibere/ca3)b 90th Pe ~90th Per ~dian centile Hedian centi le Reference 0.7 1.2 0.00002 0.00011 19 31 187 127 34 22 0.7 0.9 1.6 2.3 6.7 31.9 13.7 42.9 5.2 0.00002 0.00017 9.8 0.00003 0.00033 6.8 0.00005 0.00023 7.8 O. 00008 0.00026 0.00022 0. 00106 O. 00046 0.00143 Sebestien et al., 1980 Sebestien et al. _ _ . 1980 Coostent et at., 1982 Nicholson, 1971 U. S. Environaenta1 Protect ion Agency, 1974 Nicholson et al., 1975, 1976 Nicholson et s1., 1971 8. Ail. 0. S mile 17 22.5 82.6 0.00075 0.00275 Nicholson et al., ( O. 8 he) from 1971 asbestos spraying 9. Air in U. S. 31 16.3 72. 7 0.00054 0.00242 Constant et at., school rooms with- 1982 out asbestos 10. Air in Paris 135 1.8 32.2 0.00006 0.00107Sebastien et al. buildings with 1980 asbestos surfaces 11. Air in U.S. 28 7.9 19.1 0.00026 0.00064Nicholson et al., but ltinBs with 1975, 1976 ce~entitious asbestos 12. Air in U.S. 54 19.2 96.2 0.00064 0.00321Nicholson et al., buildings with 1975, 1976 friable asbestos 13. Air in U. S. 54 62. S 550 0.00208 0.01833 Constant et al., schoolroom with 1982 asbestos surfacce 14. Air in U.S. 27 121.5 465 0.00405 0.01550 Nicholson et al. schools with 1978 damaged asbestos surfac ing ~tcriale adopted from Nicholson, 1983. bBesed on conversion factor of 30 pg/~3 - 1 fiber/ce3.
221 fibers/cm3; for indoor uncontaminated air, it is 0.002 fibera/cm3; ant for indoor asbestos-contaminated air, it is 0.003 fibere/cm3. The same distribution of occupancy over time was used to arrive at the 0.002 fibere/cm3 figure for a high exposure level. Risk Assessments for Special Subpopulations :} .., j Table 7-2 shows lifetime risk estimates for people who are exposed throughout their liver to levels of either 0.0004 or 0.002 fibere/cm3 in ambient air. The predominant risk is from mesothelioma, but lung cancers also contribute to the rink, especially for male smokers. For exposure patterns that are different from those assumed, lifetime risks could be higher or lower. The following are three illustrations of how lifetime risks could be derived for such special populations. Children Exposed in Asbestos-Contaminated Schools. The committee - estimated the risk for persons exposed from birth to age 73 year~ to environmental levels of 0.002 fibere/cm3 (as assumed in Table 7-2) plus an additional risk from a 10-year exposure (from ages 6 to 16) in an asbestos-contaminated schoolroom for 6 hours daily, 200 days per year, to 0.02 fibers/cm3 (550 ng/m3, the 90th percentile in Table 7-6) . The equivalent continuous daily 10-year exposure is approximate ly 0.003 fibers/cm3, i.e., 0.02 x (200 x 6~/~365 x 24~. Using equation (6), the lifetime risk of lung cancer for a male who eventually becomes a smoker is 0.003 x 10 x 0.0022, or 66 in a million. This risk represents an approximately 20: addition to his ambient lifetime risk of 320 in a million (0.002 x 73 x 0.0022), for a total of about 390 in a million. For such an individual, the schoolroom exposure adds relatively more to the risk of mesothelioma, as shown below. Using equations (G4) and (G5) in Appendix G for the lifetime mesothelioma risk, L, at t ~ 73 for an exposure of Q = 10 years starting at age to a 6 at the dose level d, this risk can be calculated from the formula: L = cdil-~1 -I /(t-to)]k)(t-to~k, with d = 0.003, i" 10, t - to a 73 - 6 ~ 67, and k ~ 3.2. This lifetime mesothelioma risk becomes L = c(0.003~11-~1-~10/67~3 21~67~3 2 = 845c. If c is the median value of Table 7-4 (i.e., c ~ 2.53 x 10 8), the estimated lifetime mesothelioma risk, L, from the 10-year exposure is 21 x 10-6. This risk is then added to the background risk of 46 x 10-6 in Table 7-2, giving a lifetime mesothelioma risk for thin subpopulation of 67 x 10-6. If a million people had received such a pattern of exposures, about 67 might be expected to tie of mesothelioma. In this example, the contribution to total risk from the schoolrooms is less than that of the lifetime exposure to the lower concentrations of asbestos estimated for the ambient air. However, if the value for k in Equation (7) were higher than 3.2, the significance of the schoolroom exposures
222 would increase because of the stronger dependence on time since fire t exposure. For example, if k ~ 3.S, the highest value used in Table 7-2, the lifetime mesothelioma risk would be 910 x lo~6. If k were less than 3.2, the corresponding lifetime risk for mesothelioma would be less than 67 x lo~6. These calculations show that childhood exposures to asbestiform fibers might contribute noticeable lifetime mesothelioma risks to those so exposed. A Female Nonsmoker in a Relatively Asbestos-Free Environment. An . example of a person in a low-risk group is a female nonsmoker exposed to an average level of 0.0001 fibera/cm3. This exposure level would not be too unlikely for a person exposed primarily to rural indoor ant outdoor air, since 0.00002 fibere/cm3 is the lowest median value for all the outdoor city readings in Table 7-6. Then, the calculations in Table 7-2 would lead to a mesothelioma lifetime rick of 2.25 x 10-6 (9 x 10-6 divided by 4) plus a lung cancer lifetime risk of 0.73 x 10-6. The lifetime individual risk for such a person would be 3 x 10-6 for both types of cancer. A Male Smoker Living in an Area Contaminated with Levels of Asbei{~~;E~]b:70bs; Exposed to High Indoor Concentrations. As an ~bigh-risk person, consider an urban male smoker exposed to 0.003 fibers/cm3 for one-half the time and 0.018 fiber~/cm3 for the other half. This pattern is based on the assumption that the subject spends one-half of his time in indoor environments with a high asbestos concentration (see sample sets 13 ant 14 of Table 7-6) ant one-half either in highly contaminated outdoor environments (see sample sets 7 and 8 of Table 7-6) or in indoor environments at the high end of the distribution for rooms that are normally not contaminated with asbestos thee sample set 9 of Table 7-6~. Thus, his continuous average exposure would be approximately 0.01 fibers/cm3, i.e., 0.5~0.003) + 0.5~0.018~. m erefore, multiplying the second column of Table 7-2 by a factor of 5 (0.01 = 5 x 0.002) would give the individual lifetime risks for such a person as 1.8 x 10-3 for the two forms of cancer taken together (230 x 10-6 for mesothelioma and 1,600 x 10 ~ for lung cancer). This lifetime risk is the additional incurred risk attributable to the nonoccupational environmental exposure to asbestos and does not include the risk incurred by the smoking itself. The portion of the additional risk attributable to lung cancer is considerably higher than it would be for a nonsmoker experiencing identical asbestos exposures. COMPARATIVE RISK ASSESSMENT - Methode The goal of comparative risk assessments is to determine whether the fiber exposure in question presents risks--in terms of total number and severity of effects per year in the United States--that are about the same, considerably more, or considerably less than those assessed
- it: 7 ] 223 quant itat ive ly above . The quant itat ive as se s sment s made in the earl ie r part of this chapter were based on exposure to a generalized "asbestos" fiber. Because future exposures to asbestos in the United States will be dominated by chrysotile, rinks of lung cancer and mesothelioma from chrysotile inhalation are assumed to be approximately the same an those attributed in the quantitative assessment to "asbestos." However, if at equal doses chrysotile is less hazardous than the other kinds of asbestos, the assumption of equal potency may lead to overstated risk estimates. These comparative risks are population risks, which combine information about the inherent rinks that a given exposure to fibers could pose to an individual and information about the current and projected distribution of exposures over the U.S. population. Unlike tile quantitative risk estimates for particular assumed exposure levels, the population risk estimates can easily change along with changing patterns of production and use. Even at a known population risk level, some individuals will receive higher than average exposures and stand at correspondingly greater individual risk, whereas the majority of the population will usually have lower risks. General Method C~nci d~ratinn,: The comparative risk assessments in this chapter are based on several ~factors, such as: o f iber type - ashes tos - other fibers with some similar properties · type of effects - lung cancer - mesoehelioma · route of exposure - inhalat ion - ingestion · source of exposure · population at risk - smokers - other special groups (such as schoolchildren) 'rhe committee did not assess fibrosis or nonmalignant pleural disease because functional impairment resulting from such effects would occur much less often than would the cancers at nonoccupational levels of exposure s .
224 Taking the first three of these factors as examples, risk assessment can be visualized as a three-dimensional matrix. As shown in Figure 7-1, the best understood combinations (inhaled chrysotile and crocidolite asbestc s for lung cancer and meso~chelioma) are in the upper right "cells" of the matrix, and the less understood combinations are successively further from that position to emphasize their "distance" from the state of knowledge necessary for quantitative risk assessment. Additional ce its could be added for other combinat ions. The following combinations of fiber type, effect, and route of exposure were considered for comparative risk assessments: chry~sot i le. /gas t rointe ~ t ina 1 c anc e r/ inge s t ion chrysot i le/mesothe lioma/inge st ion crocitolite/lung cancer/inhalation crocidoliee/mesothelioma/inhalation other asbestos/all cancers/both routes fibrous glass/lung cancer/inhalation fibrous glans/mesothelioma/inhalation attapulgite/lung cancer/inhalation attapulgite/mesothelioma/inhalation mineral wool/lung cancer/inhalation mineral wool/mesothelioma/~nhalation ceramic fiber/lung cancer/inhalation ceramic fiber/mesothelioma/inhalation carbon fiber/lung cancer/inhalation carbon fiber/me sothe 1 ioma/ inha let ion The committees results are expressed in comparison with the chrysotile/lung cancer/inhalation cell, hereafter called the prime cell. Its designation as the prime cell does not imply that it is the cell corresponding to greatest population risk. According to the calculations in the preceding section, if environmental exposures to asbestos in early life are frequent, mesothelioma may prove to be the dominant effect. Both the comparative scores and the evaluation of the uncertainty in them were made qualitatively rather than quantitatively; the entries are symbols (+, O. -, a, b, c) rather than numeric. Appendix H describes how the committee went about assigning, combining, and assessing the symbolic code ~ . ~ score sheet for recording judgments about comparative risks is shown in Figure 7-2. Completed sheets for scored ceils are included in Appendix H. Thene sheets are supplied to allow the reader to evaluate the individual judgments or the con~ittee' ~ sub jective combination of them.
1 1 i Mesothelioma Lung Cancer Gastrointestinal Cancer Vet Vat -Vat Other Fibrous Other Fibers Glass Asbestos Crocidolite Chrysotile Effect Asbestos T Route fiber ~ ; FIGURE 7-1. Three-dimensional matrix for conceptualizing the risk assessment problem. j 1 COMPARATIVE RISK ASSESSMENT SCORESHEET Cell Scored / / ~; ~_ c t Route Scores Comparative with Ce 11 Fibe r Expo sure S c ore R i ad i ~ no s i t ion Score . Produc t ion Use Pattern Geography Populat ion Trends Fiber Size Morpho logy Chemistry Penetrat ion Stability / Overall risk compared with cell above Overal 1 risk compared with prime ce 11 Qual ity of comparat ive ri sk as se s sment Remarks: Effect Route Ef fee t s Score ~. Human Stud ie s Anima 1 Stud ie s In-Vitro Stud ie s Synergism Othe r FIGURE 7-2. Score sheet for recording judgment s about comparat ive risks .
226 Scoring Considerations Production. If all other factors were equivalent, a greater production volume (or U.S. consumption level, if that is significantly different) would result in a greater level of exposure and a correspondingly greater population risk. If natural occurrence is important, it can be used here as another surrogate for exposure. Use Pattern. Several concepts are embodied here. Al] have to do with the degree to which production, consumption, or natural occurrence will lead to actual human exposures. If the fibers are used only in products where they are tightly bound into a matrix, relatively little exposure will occur at least until final disposal, whereas loose fiber use in consumer applications would lead to relatively heady and immediate esposurea. Products such as talcum powder, which are intended for direct human use, will lead to higher exposures per unit production than those that are not. Geography. This a core applies to the spatial distribution of sources including natural deposits, mills or production facilities, fiber product manufacturing sites, use sites, and disposal sites. Concentrated sources tend to imply higher exposures of fewer people. This classification can also be used as a basis for evaluating such factors as the likelihood of fibers reaching drinking water. Population. The size of the population at risk determines the extent of the hazard for a given level of individual risk. A type of fiber that yields exposures to many people, such as a constituent of a common cons''=er product, has more potential for producing adverse health effects than one that affects only a few people, such as a naturally occurring but noncommercial fiber that is present only in selected, sparsely populated regions. Trends. Exposure is a dynamic process that changes with changes in total production volume, production processes, une patterns, population distribution and habits, and many other factors that do not remain static. Thus, the risk that would apply to a steady state of exposure at current levels con be misleading both for currently observed effects or for future occurrence of effects. The sharp downtrend in asbestos exposures tends to ameliorate the population risks that might otherwise be asseased, whereas new fiber types may present enormously higher exposures in the future than they do at present. Fiber Size. Two counteracting influences are at work with fiber size. The clearest is their respirability, which declines markedly as fiber diameter increases, becoming essentially zero above 3 or 4 pm. It is likely that length also eventually affects respirability and, especially, transport potential within the body. On the other hand, short fibers are probably more easily removed from the body by phagocytes; thinner ones may be more easily dissolved, coated, or gelled
227 by body fluids; and small fibers in general may not act biologically the - same as large fibers, which con disturb eany cells at once. Furthermore, am^31 fibers may be more likely to be exhaled with the tics, volume and, thus, not retained in the lung. The overall significance of fiber size may therefore be represented as a potency that is greatest for fibers around 0.2 am diameter and 20 am in length (Pott, 1978~. Morphology. Whatever the response to fiber size, it seems likely that long, thin fibers that have strength, durability, flexibility, and a high aspect ratio are more likely to cause adverse health effects than are fibers without these characteristics. The curIlness of chrysotile fiber bundles may increase their effective aerodynamic diameter, thus decreasing their respirability below that expected on the basis of fiber diameter alone. Chemistry. Although little is known about the influence of fiber chemistry on potential for health effects, it seems possible that the chemical properties of fibers play some role, especially with respect to surface chemistry. Another feature of surface chemlatry, i.e., the ability to adsorb carcinogenic substances, is included under "aynergism." Penetration. The ability of a fiber to penetrate to the site where effects are developed, for example, to the pleura or peritoneum in the development of mesothelioma, is clearly important to its potential for causing disease. Thin category includes all fiber properties that facilitate such penetration. It is closely related to fiber size, morphology, and stability. Stability. Some experimental evidence suggests that the longer a fiber remains in a tissue, the greater is its opportunity for inducing its biological effects, for example, stimulating cell hyperplasia when a transformed cell is present. In this case, the important factor is not the resistance to tranalocation but the resistance to chemical or physical degradation such as dissolution or gelling. H''=~n Studies. This category include e both clinical and epidemiological observations in Herman populations. Animal Studies. The demonatration of significant biological effects in a well-desig~ed animal experiment is considered evidence that the test substance has a potential for causing similar effects in hark. In Vitro Studies. Although the meaningfulness of short-term, in vitro experiments with respect to the effects of fibers is questionable, it is known that asbestos and some other fibers demonstrate some cellular-level effects such as hemolysis. The ability to cause such effects is considered a weak, but not entirely worthless, argument for health effects potential.
228 Synergism. Information on synergistic effects would markedly affect assessment of comparative risk. The only such information available involves asbestos and cigarette smoking. Other. This catchall category could be applied to any influence on overall risk, including exposure, biodisposition, and effects. For example, if a particular fiber is found to be more likely than the others to reach young children and if the effect in question is most prevalent in children or if it increases in incidence with time after flrat exposure as with mesothelioma, then the comparative risk estimate would be increased. Discussion of Comparative Risks Table 7-7 anm-=rizes from a different perapectlve the information in Appendis H. No cell of the fiber/effect/route matrix approaches the population risk levels associated with the prime cell (chrysotile/lung cancer/i~halation). As noted in the quantitative assesament, the mesothelloma risk from lifetime exposure to asbestos is potentially much greater than the lung cancer risk. Although some researchers question Whether chrysotile is as potent as other asbestos varieties in causing mesothelioma, the committee has assumed that even exposure ondy to chrysotile continously since birth would cause more mesothelioma then lung cancer. Chrysotile has been extensively used in the past and thus alto provider a source of in-place exposure. Of the other combinations, the commlLtee believes the ones most worth watching in the near term are fibrous glass any attapulgite for lung cancer by inhalation. The risks for effects of crocidolite and other asbestos varieties are reasonably well understood, and measures taken to reduce occupational exposures in the future may also keep the nonoccupational exposures to a mininum. However, general population exposures to crocidolite already in place could be subatantial, especially in connection with its disposal. The other cells seem to entail significantly lese population risk (more than 10 times less) than the prime cell. In several cases, this Judgment is based principalig on current exposure or biodisposition rather than on definitive evidence that the fibers have low intrinsic health effects potential. For example, both ceramic and carbon fibers can be found in Despicable size ranges and may well have biological properties similar to those of asbestos. However, they are produced in law volumes and are used in limited, generally contained applications. Population risks could become substantial if these facts ~hanged. Most fibrous glass and mineral wool is produced in nonrespirable sizes, and some evidence from epldemlological and animal studies suggests that their biological toxicity is low. Thus, risk levels for these substances are rated low despite the substantial potential for exposure.
. Factor Higher Siailar Lo~er 229 T2UL~ 7-~. Sw~ry of Coaperative ai~ Aseeseacot Coopared trith Chrycotile/Lung Cancer/~helation, Dats on the Factor Suesest that Populat ion Rick Should bc Much Lover Production Fibrous glese Minere1 wool Crocidolite Attepulgite Other ashestos Carbon f iber Ceramic f iber Use pattern Fibrous glass Other asbestos Crocidolite Cesemic fiber "tepulgite Carbon fiber Minere1 vool Cnrysotile/ingest ion Geogrephy Fibrous glass Other asbestos Crocidolite Minere1 wool At tepulgite Carbon fiber Ceremic fiber Cnrysot ile/ ingest ion Population Fibrous glass Crocidolite Carbon fiber Attapulgitc Other sabestos Ceramic fiber Minere1 vool Trend e Fibrous glase Other ashestos Crocidolite At tepu lgi te Mine re 1 woo 1 Carbon f ibe r Ceramic fiber Fiber sisc Crocidolite Minere1 vool Fibrous glese Other asbestos Attapulgite Carbon fiber Ceremic f iber Morphology Croc ido 1 ite t`1 1 others Chemistry No clear effect of cheeistry evident Penetration Crocidolite Cerbon fiber Minera1 vool Fibrous glass Other asbesto. Ceremic fiber Chrlrsotile/ingestion Atespulgiee Stabilit~r Crocidolite All otbere Fibrous g1~e Other asbesto. (continuet on next pege) - ..
"B~ 7-7 (COSIt. ) 230 Compared trite Cbrysotile/L~ng Cancer/Inb~a,~eion, Beta on the Factor Sumest that Population Blak Should be Factor ~br ~Siallar Lower Much I~wer Epideeiological Crocldolite/ Crocidolite/ Fibrous glass studies aesothelio~a lung cancer Ceramic fiber Mineral wool ~ners1 wool "i - 1 studies Crocidolite 4~, others Other asbestos In vitro studiesa Synergism Otherb Overall popt't A tion rifle aQuantltati~re differences in activity not apparent. No other factor ~s sufficiently striking for l~clualon. All others Fibrous Gil 88 Chrysotlle/ ~sothelioma/ ingestion Crocidolite Attapulgitc/ lung cancer Fibrous glass Carbon fiber Ceramic fiber At~pulgite/ aesothelioo~a Other asbestos/ other cancer For any combination of fiber type, effect, and route of exposure not assessed, even for comparative risk, the committee believes either that risks are at most of marginal significance or that there is insufficient information on which to base such a comparison. Most of the combinations fall into the former category. Carcinogenic effects other than Burg cancer or mesothelioma constitute examples of the insufficient information category for several fibers. SUMMARY AND RECOMMENDATIONS The committee has made quantitative risk assessments for nonoccupational exposures to asbestos and qualitative (or comparative) risk assessments for a variety of asbestiform fibers. Lung cancer and mesothelioma from inhaled materials received the greatest consideration.
231 l ] .a .. i . i .i For the quantitative risk assessment, a linear model for low dose extrapolation was used. When quantifying risk from nonoccupational exposures, uncertainties are introduced not only by the selection of mathematical models but also because the characteristics of fibrous materials in the ambient environment differ from those in the workplace. By converting mass concentrations measured in the environment to equivalent numbers of fibers in the workplace, the committee assumed a median population exposure of 0.0004 fibers/cm3 air throughout a 73-year lifetime. Based on this and various other assumptions, the individual lifetime risk for lung cancer was estimated to be between 3 in a million for female nonsmokers and 64 in a million for male smokers, and for me~othelioma it was approximately nine in a million, regardless of smoking habits or sex. However, other assumptions could decrease the risks essentially to zero, or could increase them. The finding that the risk for mesotheliama is greater than that for lung cancer among nonsmokers is due to the strong dependence of mesothelioma risk on time since first exposure. Thus, a given exposure in childhood markedly increases the lifetime risk of mesothelioma compared with an equivalent dose later. It should be remembered that these risk estimates were based on data obtained from worker cohorts. Smokers runs a substantially higher risk of malignant disease from asbestos than do nonsmokers; for smokers, lung cancer is a greater risk than mesothelioma. Studies should be conducted to learn more precisely the dependence of mesothelioma and lung cancer mortality on time since first exposure and on the characteristics of the exposure. Such efforts should include studies in animal models and follow-up studies of occupationally exposed cohorts. For the comparative risk assessment, population risks (as opposed to individual risks) were considered. The risks were based on three major factors: exposure levels, biodi~position, and evidence of adverse health effects. m e potential for exposure was a dominant factor. Emus, risk estimates for substances of equal biological potency may be widely divergent if the populations exposed to them differ greatly. Two points follow from this. First, some individuals may be exposed to high levels of a fiber for which the overall population exposure is low. Second, the overall population risk would change if use patterns change. Current population risk from exposures to the various substances considered, including fibrous glass, attapulgite, and carbon fibers, appears to be much less than for the risk from asbestos, especially chrysotile. However, further information is needed to evaluate the possible adverse effects of exposures to fine fibrous glass and attapulgite.
232 REFERENCES Armitage, P. 1982. The assesament of low-dose carcinogenicity. Biometrics 38~supplement):119-129. Armitage, P., and R. Doll. 1961. Stochastic models for carcinogenesis. Pp. 19-38 in Proceedings of the Fourth Berkeley Symposium on Statistics and Probability. Yol. 4. University of California Press, Berke ley . Babich, H. , and D. L. Davis. 1981. Food tolerances and action levels: Do they adequately protect children? Bioscience 31:429-438. Calabrese, E. J. 1978. Pollutants and High-Risk Groups. Wiley- Interacience, New York. Calabrese, E. J. 1980. Nutrition and Environmental Health. Vol. 1: me Vitamins. Wiley-IDteracience, New York. Constant, Jr., P. C., F. J. Bergman, and G. R. Atkinson. 1982. Airborne asbestos levels in Schools. Final Report, Environmental Protection Agency. Midwest Research Institute. Contract 68-01-5915. Consumer Product Safety Commission. 1983. Report by the Chronic Hazard Advisory Panel on Asbestos. Consumer Product Safety Commission, Washington, D.C. Cornfield, J., F. Carlborg, and J. Van Ryzin. 1978. Setting tolerances on the basis of mathematical treatment of lose-response data extrapolated to low doses. Pp. 143-164 in G. L. Plea and W. A. M. Duncan, eds. Proceedings of the First International Congress on Toxicology. Academic Press, New York. Crump, K. S., D. G. Heel, C. H. Langley, and R. Peto. 1976. Fundamental carcinogenic processes and their implications for low dose risk assessments Cancer Res. 36:2973-2979. Dement, J. J., R. L. Harris, M. J. Symons, and C. Shy. 1982. Estimate o f los e-respons e for respiratory cance r among chrysoti le ashes toe textile workers. Ann. Occup. Hyg. 26: 869-887. Doull, J., C. D. Klaa~sen, and M. 0. "cur, ede. 1980. Casarett and Doull's Toxicology: The Basic Science of Poisons. Macmillan, New York. 778 pp. Enterline, P. E. 1983. Cancer produced by nonoccupational asbestos exposure in the Udited States . J. Air Pot lut . Control Assoc . 33: 318-322. Fishbein, L. 1980. Overview of some aspects of quantitative risk assessment. J. Toxicol. Environ. Health 6: 1275-1296. Food Safety Council. 1980. Proposed System for Food Safety Assessment. Food Safety Council, Washington, D.C. Golder~thal, E. I. 1971. A compilation of LD,o values in newborn and adul t anima ~ ~ . Toxico l . App l . Pharmaco l . 18 :185-20 7 . Hammond, E. C., I. J. Selikoff, and H. Seidman. 1979. Asbestos exposure, cigarette smoking and death rates. Ann. N. Y. Acad. Sci. 330:473-490. Hartley, H. 0., and R. Sielken. 1977. Estimation of "safe doses" in carcinogenesi~ experiments . Biometrics 33: 1-30. Hartley, H. 0., H. Tolley, and R. Sielken. 1981. The product form of the hazard rate model in carcinogenic testing. Pp. 185-200 in M. Csorgo, D. A. Dawson, J. N. K. Rao, and E. Saleh, eds. Statistics and Related Topical North-Holland, Amsterdam. Henderson, tl., and P. E. Enterline. 1979. Asbestos exposure: Factors associated with excess cancer and respiratory mortalilty. kin. N.Y. Acad. Sci. 330:117-126.
'1 233 Hobbs, H. S. T., S. D. Woodward, B. Murphy, A. W. Musk, and J. E. Elder. 1980. The incidence of pneumoconio~i., mesothelioma and other respiratory cancer in melt engaged in mining and milling crocidolite in tiestern Australia. Pp. 615-625 in J. C. Wagner, ed. Biological Effects of Mineral Fibres. IARC Scientific Publication No. 30. International Agency for Research on Cancer, Lyon. Hoel, D. G. 1980. Incorporation of background response in dose-response mode Is. Fed. Proc . 39: 73-75. Heel, D. G., N. L. Kaplan, and M. W. Anderson. 1983. Implication of non-linear kinetics on risk estimation in carcinogenesis. Science 219:1037. Interagency Regulatory Liaison Group. 1979. Work Group on Risk Assessment. Scientific bases for identification of potential carcinogens and estimation of risks. J. Natl. Cancer Inst. 63:242. Kalbfleisch, J. D., and R. L. Prentice. 1980. The Statistical Analysis of Failure Time Data. John Wiley and Sons, New York. Kalbfleisch, J. D., D. Krewski, and J. Van Ryzin. 1983. Dose response models for time to response toxicity data. Can. J. Statist. 1 1: 25-4f'. Klasssen, C. D., and J. Doull. 1980. Evaluation of safety: Toxicologic evaluation. Chapter 2 in J. Doull, C. D. Klasssen, and M. 0. Amdur, eds. Casarett and Doull' ~ Toxicology: The Basic Science of Po i sons . Macmi 1 1 an, New York . Krewski, D., and J. Van Ryzin. 1981. Dose response models for quantal response toxicity data. Pp. 201-231 in M. Csorgo, D. A. Dawson, J. N. K. Rao, and E. Saleh, eds. Statistics and Related Topics. North-Holland, Amsterdam. McDonald, J. C., and F. D. K. Liddel1. 1979. Mortality in Canadian miners and millers exposed to chrysotile. Ann. N. Y. Acad. Sci. 330:1-9. National Research Council. 1977. Principles and Procedures for Evaluating the Toxicity of Household Substances. A report of bile Committee for the Revision of NAS Publication 1138, Assembly of Life Sciences. National Academy of Sciences, Washington, O.C. 130 pp. National Research Council. 1983. Risk Assessment in the Federal Gove rnment: Manag ing the Proce ~ ~ . A report of the Commit tee on the Institutional Means for Assessment of Risks to Public Health, Commission on Life Sciences. National Academy Press, Washington, D. C. 191 pp. Newhouse, M. L., and G. Berry. 1976. Predictions of mortality from mesothelial tumours in asbestos factory workers. Br. J. Indus. Med. 33: 147-151. Newhouse, M. C., and G. Berry. 1979. Patterns of disease among long-term asbestos workers in the United Kingdom. Ann. N. Y. Acad. Sci. - 330: 53-60. Nicholson, W. J. 1971. Measurement of asbestos in ambient air. National Air Pollution Control Administration. Final Report, Contract CPA 70-92.
234 Nicholson, W. J. 1978. Control of sprayed asbestos surfaces in school buildings: A feasibility study. Final Report. Contract I-ES-2113. National Institute of Environmental Health Sciences, Research Triangle Park, N. C. Nicholson, W. J. 1983. Health Ef feces Update . June 1983. Unpublished draft prepared by W. J. Nicholson, Mt. Sinai School of Medicine, City University of New York. 148 pp. Nicholson, W. J., A. N. Rohl, and E. F. Ferrand. 1971. Asbestos air pollution in New York City. Pp. 136-139 in H. M. England and W. T. Barry, eds. Proceedings of the Second Clean Air Congres s . Academic Press, New York. Nicholson, W. J., A. N. Rohl, and I. Weisman. 1975. Asbestos contamina- tion of tile air in public buildings. EPA-450/3-76-004. Environmental Protection Agency, Research Triangle Park, N.C. Nicholson, W. I., A. N. Rohl , and I. Weisman. 1976. Asbestos contamina- tion of building air supply systems. Paper 29-6 in Proceedings of the International Conference on Environmental Sensing and Assessment. Institute of Electrical and Electronic Engineers, Vol. II. Institute of Electrical and Electronic Engineers, New York. Nicholson, W. J., I. J. Selikoff, H. Seidman, R. Lilis, and P. Formby. 1979. Long-term mortality experience of chrysotile miners and millers in Thetford mines, Quebec. Ann. N. Y. Acad. Sci. 330: 11-21. Nicholson, W. J., A. N. Rohl, I. Weisman, and I. J. Selikoff. 1980. Environmental asbestos concentrations in the United States. Pp. 823-827 in J. E. Wagner, ed. Biological Effects of Mineral Fibres. Vol. 2. IARC Scientific Publication No. 30. International Agency for Research on Cancer, Lyon. Of fice of Technology Assessment. 1981. Assessment of Technologies for Determining Cancer Risks from the Environment. Office of Technology Assessment, Washington, D. C. 240 pp. Pe lkonen, ()., E. H. Kalt iala , T. K. Karmi , and N. T. Kirks . 197 3 . Comparison of act ivities of drug-metabolizing enzymes in human fe tat and adult livers. Clin. Pharmacol. Ther. 14: 840-846. Peto, R. 1978. Carcinogenic effects of chronic exposure to very low levels of toxic substances. Environ. Health Perspect. 22: 155-159. Peso, J. 1980a. Lung cancer mortality in relation to measured dust level s in an asbestos textile factory. Pp. 829-836 in J. C. Wagner, ed. Biological Ef fiche of Mineral Fibres. Vol. 2. I\RC Scientif ic Publication No. 30. International Agency for Research on Cancer, Lyon. Peto, J. 1980b. The incidence of pleural mesothelioma in cl~rysotile asbestos textile workers. Pp. 703-711 in J. C. Wagner, ed. Biological Effects of Mineral Fibbed. Vol. 2. lARC Scientific Publication No. 30. International Agency for Research on Cancer, Lyon. Peto, J. 1982. Dose and time relationships for lung cancer and meso- thelioma in relation to smoking and asbestos exposure. Presented at the Symposium on Asbestos Carcinogenesis. Feb. 17-19, 1982, in West Berlin. Organized by Bundesgesundheits~mt.
235 Peto, J., H. Sei~n, and I. J. Selikoff. 1982. Mesothelioma mortality in asbestos workers: Implication for models of carcinogenesis and rink assesament. Br. J. C:'ancer 45:124-135. Pott, F. 1978. Some aspects or the dosimetry of the carcinogenic potency of asbestos and other fibrous dusts. Staub Reinbalt. Lift 38: 486-490. Rane , A., and E. Ackerman. 1972. aniline in human fetal liver. Metabolism of ethylmorphine and Clin. Pharmacol . men 13: 663-670. Richmond, C. R., P. J. Walsh, and E. D. Copenhaver, edit. 1981. Proceedings of the bird Life Sciences Symposium, Health Risk Analysis , Ga tlinburg , Tent ., Oct . 27-30, 1980. Franklin Ins titute Pren ~ . Philadelphia . 438 pp . Schneiderman, M. A., I. C. Nisbet, and S. M. Brett. 1981. Assessment of risks posed by exposure to low levels of asbestos in the general environment. Prepared for Institut fur Wasser, Boden, und L~fthygiene des Bundeagesundheitsamtes. No. 4. Dietrich Reimer Verlag, Berl in. Sebastien, P., M. A. Billion~Galland, G. Dufour, and J. Bignon. 1980. Measurement of asbestos air pollution inside buildings sprayed with asbestos. EPA 560/13~0~026. Environmental Protection Agency, Washington, D. C. Seidman, H., I. J. Selikoff, and E. C. Hammond. asbestos work exposure and long-term observation. Sci. 330: 61-89. , 1979. Short-term Ann . ~ . Y . Aca d . Selikoff, I. J., E. C. Raymond, and H. Seidman. 1979. Mortality experience of insulation workers in the U.S. and Canada, 1943-1976. Ann. N.Y. Acad. Sci. 330:91-~16. Tager, I. B., S. T. Weiss, A. Munoz, B. Rodney, arid F. E. Speizer. 1933. Longitudinal study of the effects of maternal smoking on pulmonary function in children. N. Engl. J. Med. 309:699-703. U.S. Environmental Protection Agency. 1974. A preliminary report on , asbestos in the Duluth, Minnesota area. Office of Technical Analysis. Environmental Protection Agency, Washington, D.C. Van Ryzin, J. 1980. Quantitative risk assessment. J. Occup. Med. 22:321-326. Van Ryzin, J. 1931. ca rcinogenicity . Discussion: The assessment of low~dose Biometrics 38 (suppl. 2) :130-139.
