JIflOIIRIIITRL~PO11CY ANALYSIS
Is There a Safe level of Exposure to a Carcinogen? STEVE E . HRUDEY Department of Public Health Sciences, Faculty of Medicine, UniversityofAlberta Edmonton, Alberta T6G 2G3 Canada
DANIEL KREWSKI Environmental Health Directorate.Healrk Protection Branch, Health Canada Omwa, Ontario KIA OL2 Canada
Estimating the contribution of environmental contaminants to the human cancer burden has been a contentious issue in environmental risk management. Low-level carcinogen exposures have been particularly controversial. Most people believe there is no safe level of carcinogen exposure, a view shared by one in five toxicologists. This suggests that extrapolating downward from high exposures in search of a de minimis cancer risk level has not provided a universal answer t o the title question. In search of a rational answer, w e present an approach of working upward from the smallest conceivable chronic dose. Using conservative assumptions, calculating the risk from lifetime exposure t o one molecule a day of the most potent known carcinogen shows that exposing the entire world population to the smallest nonzero dose likely would not yield a single case of cancer. Less potent carcinogens present correspondingly lower risk. These calculations suggest that, within a realistic concept of safety, there is a safe level of exposure. This simple analysis applies the no-threshold cancer risk estimation methods used by EPA and provides a rationale t o answer the title question. Ultimately, this perspective may open dialogue on the more difficult question of how much carcinogen exposure could be considered safe.
370 A
VOL. 29. NO. 8.1995 I ENVIRONMENTAL SCIENCE 5 TECHNOLOGY
The recognition that chemical substances in the environment can cause cancer in humans has dominated the environmental agenda since the 1970s. Nearly one-third of all North Americans will contract cancer during their lifetimes and about onequarter will die of some form of this group of dreaded diseases. Thus, public concern about the causes of cancer is understandable. However, the role of environmental exposure to carcinogens as factors in the causation of cancer has been controversial. A 1964 World Health Organization report coucluded that three-quarters of all human cancers are caused by extrinsic factors [causes other than genetic predisposition) (1). Extrinsic includes the whole range of environmental factors involved in bring, including but not limited to exposure to chemical carcinogens. Yet, the view that chemicals made by humans are responsible for the majority of human cancers became popular in the 1970s with the aid of such literature as Epstein’s book The Politics of Cancer. This view remains common: A survey of Pennsylvania adults found one in five respondents believed that between 25% and 50% of all new cases of cancer were caused by toxic chemicals from waste sites (2). In contrast, a 1981 scientific review estimated that perhaps only 2%-within a range from < 1% to 5 M f human cancers arise from pollution sources, as distinct from tobacco, alcohol, diet, food additives, occupation,indusnial products, medicines, and geophysical factors (3).This finding was consistent with a 1989 analysis that found a summation of all EPA upper bound cancer risk assessment values for “pollution”causes would account for between 1%and 3% of all human cancer cases in the United States (4).
Regardless of the portion of total cancer risk that can be directly attributed to environmental contaminant exposures, cancer causation remains a primary concern that is likely to occupy a prominent place in the environmental agenda for the foreseeable future. As much consensus as possible must be found among the divergent views, particularly with respect to low levels of exposure to substances with carcinogenic potential. First, we should ask Is there a safe level of exposure to a carcinogen?
Expert and public opinion Given the advancement of knowledge about cancer resulting from expending billions of dollars on biomedical research, the public has reasonable grounds to expect that the scientific community could provide a clear and unequivocal answer to such an apparently fundamental question. However, a survey found substantial divergence of responses among members of the U.S. Society of Toxicology to the SUT0013-936x/95/0929-370A$09.00/0 0 1995 American Chemical Swiety
vey statement: “There is no safe level of exposure to a cancer-causing agent” (5). Among toxicologists, 18.7% either agreed or svongly agreed that there is no safe level Of sure;another6.6%rewondedthat they did not know or had no ooinion. In a comoarison ~ O U Qof Portland, OR, ckzens, more t h i 50% agreed with the assertion that there is no safe level of exposure and 11.3% did not know or had no opinion. Similar answers to comparable questions recently have been reported for members of the Canadian public (6)and the Society of Toxicology of Canada (7). The fact that 75% of toxicologists disagreed with the survey statement might normally be regarded as a substantial level of scientific consensus, were it not for the apparently fundamental nature of this question and the clear divergence of the public new from the majority of the experts. Considerable divergence of opinion also occurred among experts, grouped in terms of employment. Thirtyfour percent of academic toxicologists agreed that there is no safe dose compared with 19% of regulatory toxicologists and only 5% of industry toxicologists. Much of the disagreement among toxicologists likely is related to the wording of the question, wbich asks about a “safe”level of exposure. What an individual considers safe involves some degree of value judgment (8). In this case, the different groups’ values apparently influenced their judgment. This implied controversy among scientists concerning the assessment of carcinogen risk likely will be interpreted by the public as chaos among the exp e w and, unless rational bounds can be placed upon the apparent disagreement,will lead to increased anxiety, distrust, and ditsculty in risk communication (9). Arguments based on a belief that there is no safe dose for a cardnogen have prevailed and have driven public policy (10). Other differencesof opinion among experts and the public may also be cited. Ames and Gold argue that fruits and vegetables contain natural pestiudes that exhibit carcinogenicpotential, and the carcinogenic risks of these may be greater than those of syntheticpesticide residues in food (11). Of the Sodeiy ofToxicology of Canada members, 73.3% agreed with the statement: “FNitS and vegetables contain natural substances that can cause cancer.” Only25.6% of the Canadian public supported this view (6, 7). About 88% of the experts disagreed with the statement: “Naturalchemicals are not as harmful as manmade chemicals,” but 56.1% of the public strongly agreed with it. On the question of safety, the survey stated, “I believe that a risk-free environment is an attainable goal for Canada.” Some 61% of the Canadian public agreed with the statement, whereas 78% of Canadian toxicologists surveyed disagreed. This
- .
,
with permission.AFB1 isaflatoxin B1,TCOD is2,3,7,8tetrachlorodibenzo-p dioxin. Potency values are not based on q,“ but on an approximation (62) obtained by linear extrapolation from TO, values 163 - -
L
’‘\ / I I/
U’
I
104
104 PC
2
102
l(r
suggests that public expectations for safety exceed what realistically can be achieved.
The linear, no-threshold kypotbesis Quantitative cancer risk assessment, in large part, is based on an evaluation of the dose-response (risk) curve. The calculation of cancer risk at e m m e l y low doses has attracted considerable discussion and controversy (12-16). The premise underlying linear extrapolation of risk to zero carcinogen dose is that a single molecule of a DNA-reactive carcinogen can damage a single DNA molecule. Furthermore, singular DNA damage can multiply through cell replication from an insigniscant molecular anomaly into a population of damaged clone cells that ultimately may become a malignant tumor. This outcome dearly is not certain; everyone has been exposed to countless carcinogens, yet not everyone develops cancer. For example, up to 90% of lung cancer is attributable to tobacco smoking (17,18). but only a minority of smokers ever develop lung cancer. Cancer development involves a number of stages, influenced by one or more contributing causal factors (19-21). Detoxification and repair mechanisms defend the body against adverse toxic effectsincluding carcinogenesis (22).Recent advances in understanding the body’s DNA repair capacity suggest that cancer occurrence may be related more to the failure of DNA repair capability than to the degree of trace exposures to DNA-damagingcontaminants (23). Even dowing for repair of pre-cancerous lesions, the VOL. 29. NO. 8,1995 I ENVIRONMENTALSCIENCE
a TECHNOLOGY
a71 A
linear no-threshold hypothesis holds that there is some nonzero probability of a singular molecularlevel event developing into a tumor. This concept of the probability of an adverse outcome following a complex sequence of stochastic events is analogous to infectious disease transmission. Some probability, but not certainty, exists that exposure to a single biological agent, which can replicate itself-a viable, pathogenic bacterium or viruswill generate sufficient numbers of “offspring”to ultimately cause disease. The probability of developing the active disease will be determined by the likelihood that the replicating pathogen can survive the host’s immune defense. Several theoretical arguments have been forwarded supporting the hypothesis of low-dose linearity for chemical carcinogens. Mathematical carcinogenesis models such as the Armitage-Doll multistage model (24) or the Moolgavkar-VenzonKnudson two-stage clonal expansion model (25) predict linearity at low doses when mutation rates associated with one or more stages relate linearly to dose. Low-dose linearity should hold whenever tumors are produced by an acceleration of an ongoing background process in a dosewise additive fashion (26, 27). Empirical evidence supporting low-dose linearity is difficult to obtain because the low risk levels associated with low exposure levels are outside of the directly observable response range (28).The US. National Center for Toxicological Research ED,, study, done in the 1970s, involved more than 24,000 mice and was designed to define the dose-response curve for the experimental carcinogen 2-acetylaminofluorine (2-AAF)down to a one-in-100 excess risk level ( 2 9 ) . Extrapolation of 2-AAF data below this point proved inconclusive (30, 3 1 ) . A subsequent largescale rodent bioassay of several nitroso compounds conducted by the British Industrial and Biological Research Association also failed to define the shape of the dose-response curve at very low doses (32).
Large-scale epidemiological investigations also are unlikely to resolve uncertainties about low dose risks. Uranium miners exposed to high concentrations of radon gas before the introduction of improved ventilation techniques were at increased risk of lung cancer ( 3 3 ) , but large-scale case control studies have failed to clarify the level of risk associated with much lower radon concentrations in homes ( 3 4 ) .With the possible exception of chronic lymphocytic leukemia, combined data analysis on more than 96,000 nuclear workers from several countries failed to establish a clear cancer risk associated with low levels of exposure to ionizing radiation ( 3 5 ) . Recent advances in molecular biology may provide some information on the shape of the doseresponse curve for DNA-reactive carcinogens at low doses. DNA adducts have been observed to be related linearly to dose even at levels much lower than those studied in the large-scale animal cancer tests ( 3 6 - 3 8 ) . Mutation frequency also has been observed to be related linearly to dose below cytotoxic levels in the Ames salmonellaimicrosome assay ( 3 9 ) .But the relevance of these observations to human cancer risk is not clear in light of detoxifica372 A
VOL. 29. NO. 8, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
tion and other protective mechanisms that may prevent the reactive metabolite from entering the nucleus or even reaching the target tissue. Many carcinogens require metabolic activation before posing a cancer risk. Saturating activation or detoxification pathways can lead to a nonlinear doseresponse curve, such as occurs with methylene chloride when the glutathione-S-transferase pathway is saturated (40).Physiologically based pharmacokinetic models now are used commonly to predict biologically relevant tissue doses (41)and afford an opportunity for more accurate cancer risk estimates, particularly when extrapolating from animals to humans. However, many nonlinear kinetic processes, such as Michaelis-Menten kinetics, essentially are linear at low doses and thus are compatible with the low-dose linearity hypothesis. Arguments may be made against the hypothesis of low-dose linearity and zero intercept for nonDNA-reactive carcinogens. Ames and Gold noted that certain chemicals may increase cancer risks indirectly by stimulating cell division, increasing the possibility of DNA damage during mitosis at high doses but not necessarily at low doses ( 4 2 ) .The International Agency for Research on Cancer recently prepared a detailed review of the mechanisms by which carcinogens act ( 2 0 ) . More than 50% of the chemicals tested under the U.S. National Cancer Institute’s National Toxicology Program during the past two decades were observed to cause tumors in rats or mice subjected to high exposure levels over the course of a lifetime (43). Quantitative estimates of cancer risk are highly correlated with the maximum tolerated dose (MTD)used in such experiments ( 4 4 ) ,which raises questions about the suitability of current protocols of carcinogen evaluation (45).Gaylor suggested exploiting this correlation to obtain preliminary risk estimates based on the MTD ( 4 6 ) .A recent U S . National Research Council report on this issue gives a detailed discussion of this point ( 4 7 ) . It is not yet possible to determine with certainty whether a threshold for chemical carcinogenesis exists. Faced with this uncertainty, regulatory agencies have entertained the hypothesis of low-dose linearity with a zero intercept as a prudent default assumption in the absence of evidence to the contrary (48, 49).
Seeking a definition of safety Although the uncertainty about low-dose cancer risks is not easily resolved, it may still be possible to develop a reasoned response to the dichotomous question: Is there a safe level of exposure to a carcinogen? To answer this question, we must consider what constitutes a safe dose. Much of the safety and risk debate concerning carcinogens has been directed at policies seeking a de minimis risk. The value of onein-a-million lifetime risk as a definition of de minimis or acceptable risk has achieved virtual folklore status in the United States ( 5 0 ) . However, regulatory applications have been less precise, with a range of risk levels now generally preferred to a single risk number (51). Most risk analysts can easily convince them-
Estimates of risk for the smallest indivisible daily dose of a carcinogen Molecular weight Carcinogen
2.3.7.8-TCDD
Benzolalpyrene Vinyl chloride Benzene
shot 322 252 62.5 78
1 molecule per day dose equivalent, d ImWWdavI
Potency
factor 9,‘
Upper bound lifstime cancer risk estinmte. r
lmgllrdldavt’
0.9 i r ~ 1.56 105 0.71 x lUm 5.8 1.9 0.18~ lrz0 0.22 Y 1 C P 2.9 x 1WZ
1 . 4 1045 ~ 4.1 x lWz0 3.3 x 1rr*’ 6.3 1rz3
Global lifnime population cancer risk (number d cases of cancer)
1 x lo* 3 x 1U’O 2 Y 10.” 4 1~13
23.1.8-TCDD is included in this snaksis only because its potency q3’ exceeds that of other known carcinogens by a large margin.
selves that lifetime one-in-a-million risk is truly de minimis, given that everyone has a cumulative risk of death that ultimately must equal one and the annual risk of death for most of one’s life is about one in 1000. In comparison, many will spend considerable money on lottery tickets that offer much lower odds, given the certainty that somebody will win. Of course, in the carcinogen lottery, the prize is not one that we wish to “win” for ourselves or our loved ones. But it is not surprising to find many who do not consider a de minimis standard of one-in-a-million lifetime risk to be safe. The negative reaction to the implied value judgment of de minimis caused regulatory agencies to abandon the term “virtuallysafe dose” for carcinogens in favor of the more nondescript “risk-specific dose” (RSD) (52).Although RSD is defined as the dose corresponding to a specified, presumably low risk level, it carries no connotations about the acceptability of that risk Nor does it necessarily cany clear indications of safety. Although de minimis risk has been implemented explicitly or implicitly for a variety of regulations on exposure to potentially carcinogenic substances, the debate over what constitutes a de minimis risk has not resolved, for many people, whether a safe carcinogen dose is possible. Perhaps we need a pragmatic expression of safety like that articulated by Yukon First Nation leader Malcolm Dawson, who said, “a safe level is one that you do not need to worry about” (53).This statement provides another way of expressing de minimis risk, which may assist in discerning what level of information is required to make a personal decision about safety. Whether any citizen or scientist will worry about a given cancer risk obviously is a personal choice with implicit value judgments. However, few in our society have been provided with the information, in understandableform, to enable them to judge this question for themselves. The conventional approach to quantitative cancer risk assessment has been to extrapolate downwards toward lower doses to calculate correspondingly small risk levels However, to find whether there is a dose level that we need not worry about, a more useful approach is to work upwards from the smallest conceivable dose that could be experienced.
be done using diiierent statistical techniques. Cnunp used an upper confidence limit (9,’) on the linear term in the multistage model as the basis for lowdose risk assessment in accordance with what has come to be known as the “linearized multistage” (Lh4S) model (54). Krewski et al. proposed a modelfree method of low-dose extrapolation (MIX) that invokes the assumption of low-dose linearity and zero intercept, but makes no further assumptions about the underlying dose-response model (55). Gaylor observed that linear extrapolation will provide an upper bound on risk should the dose-response curve he nonlinear and curve upwards at low doses (56). The EPA approach to calculating the RSD usually is based on Crump’s LMS model (which is generally equivalent to MFX) and is explained as providing an upper bound estimate of cancer risk. Accordingly, we can ask what this method, generally accepted as conservative, will predict as the risk for the smallest indivisible daily dose of a carcinogen. EPA-calculated RSDs are based on chronic lifetime exposure. Hence, the appropriate value for the smallest indivisible chronic daily dose would be one molecule of carcinogen per day for an entire lifetime. However, the risk for a one-time exposure to a single molecule would be 25,550 (i.e., 365 days/
Estimating cancer risks At the technical level, extrapolating doseresponse data on carcinogenesis from high to low dose may
VOL. W , NO. 8.1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY rn 11711 A
year x 70 years) times lower than the calculation we present below. Calculating the corresponding risk requires simple arithmetic based on the formula r = q,* x d. Here, risk (r) is the probability of cancer occurrence associated with exposure to a daily dose (d) for life (say 70 years). To illustrate, consider the case of 2,3,7,8 tetrachlorodibenzo-p-dioxin (TCDD).A dose of one molecule per day translates into mass by dividing the molar mass (322,000 mg) by Avogadro's number (6.02 x loz3molecules/mole) and by a lifetime average body mass of 60 kg (57)to yield an equivalent dose of 0.9 x lo-'' mg/kg per day. This dose is multiplied by the q,* value of 1.56 x lo5 to yield a lifetime cancer risk Similar calculations for a few selected of 1.4 x carcinogens with published ql* values are summarized in Table 1 (58-60). What conclusions can be drawn from these calculations? First, it is worth noting the enormous range of chemical carcinogen potencies. As illustrated in Figure 1, these cancer potency values vary by more than 10 million-fold (61, 62). The potency values shown in Figure 1 are not based on ql*, but rather on an approximation obtained by linear extrapolation from the TD50 values (63,64). By itself, this enormous range suggests that exposure to low-potency carcinogens will be inherently less dangerous than exposure to high-potency carcinogens. Because it has the highest reported potency by a substantial margin, TCDD was included in our analysis even though it is not DNA reactive. Although TCDD may not be genotoxic, EPA failed to rule out the possibility of lowdose linearity in its recent reassessment of TCDD health risk (65). The probability of developing cancer from exposure to one molecule of TCDD a day over a lifetime is calculated by EPA's upper bound approach as less than 2 x Given the astronomical numbers of cells and molecules in the human body and the myriad factors that could prevent the interaction of a carcinogenic molecule with a strand of DNA from ultimately creating a tumor, it is not surprising that we should calculate an extremely small risk for such an infinitesimally small exposure. For individuals to decide whether a safe level exists, exploring the dimensions of this calculated risk for the smallest possible indivisible daily dose is instructive. Using the calculated risk value and assuming the same lifetime exposure to TCDD for the entire world population indicates that there would be a less than 1 in 100,000 chance of even a single cancer case arising from that exposure level. From another perspective, the calculated risk value indicates that all humans who have ever lived on the planet could have been exposed to this infinitesimal level without producing a single cancer case in human history. This analysis does not attempt to consider the current exposure levels to TCDD nor to evaluate what risk levels may be associated with those exposure levels because different mechanisms likely apply. Furthermore, this calculation for a hypothetical infinitesimal exposure neither supports nor refutes arguments on the contentious issues surrounding the human health significance of TCDD as an environ374 A
VOL. 2!9, NO. 8, 1995 /ENVIRONMENTAL SCIENCE & TECHNOLOGY
mental contaminant (65). However, it does illustrate that there is a quantifiable level of exposure, which most individuals likely would consider below their worry threshold. The TCDD example involves the highest reported potency of any known carcinogen; similar calculations for other carcinogens yield substantially lower risk levels. To further illustrate the potency range of carcinogens according to their ql' values, consider the plots in Figure 2 of the dose-response relationships, using an arithmetic scale, for the carcinogens summarized in Table 1. This depiction shows the magnitude of divergence of the doseresponse relationships for individual carcinogens because of differing potencies.
Conclusions This highly simplified analysis clearly suggests that, without great philosophical difficulty, an individual can reconcile the existence of a safe exposure level for any known carcinogen with the linear, nothreshold (zero intercept) hypothesis for carcinogen risk assessment. This reconciliation can be made without endorsing or disputing the merits of current regulatory policy, which favors this hypothesis, at least for DNA-reactive carcinogens. Regardless of the merits of any specific exposure value, there is merit in reconciling the boundaries of concern about carcinogens. The premise that there are no safe exposure levels to a carcinogen has profound implications. As noted previously, a majority of the Canadian public believes that a risk-free environment is attainable. To achieve a zero-risk environment would require total elimination of exposure to even a single molecule of any cancercausing agent, if we accept the linear, no-threshold hypothesis. In reality this is not possible. Rather, we need an operational concept of environmental safety that is more practical than zero risk. The prevalent belief that there can be no safe level of exposure to a carcinogen because there may be no zero-risk exposure level may hinder progress in achieving effective public policy on this issue. The main point of this article is to establish that, under any realistic concept of practicality, there is indeed a safe level of exposure to cancer-causingchemicals. The essence of our argument is that lifetime daily exposure to the smallest possible amount (one molecule) of even the most potent cancer-causing chemical known to date would pose a de minimis risk by even the most cautious standards. The more difficult question of what exposure level should be tolerated for cancer-causing chemicals in the environment cannot be answered by our analysis. In practice, the degree to which exposure to environmental carcinogens should be controlled is a complex issue that requires careful consideration of risks, costs, and benefits, in accordance with the legal statutes underlying carcinogen regulation (66).We argue that an operational definition of safety, which is consistent with conservative cancer risk estimation hypotheses, can be satisfied by extremely low exposure levels above zero, thereby admitting risk management options other than complete elimination of exposure. On an individual level, interested scientists who
may be asked if there is a safe level of exposure to a carcinogencan use this analysis to develop their own clear and reasoned response. Thus they may avoid contributing to impossible expectations or unwarranted fears in our society.
Acknowledgments This work was supported by funding for the Eco-Research Chair in Environmental Risk Management provided by the Tri-Council Secretariat representing the Medical Research Council of Canada, the Natural Sciences and Engineering Research Council of Canada, and the Social Sciences and Humanities Research Council of Canada; the Alberta Heritage Foundation for Medical Research; Alberta Environmental Protection; Alberta Health; and the City of Edmonton.
References (1) “Preventionof Cancer”;Technical Report Series 276;World Health Organization: Geneva, 1964. (2) Bord, R. J.; O’Connor, R. E. Risk Anal. 1992, 12(3),41116. (3) Doll, R.; Peto, R. J. Natl. Cancer Inst. 1981, 66(6), 11931308. (4) Gough, M. Environ. Sci. Technol. 1989, 23(8),925-30. (5) Kraus, N.; Malmfors, T.; Slovic, I! Risk Anal. 1992, 12(2), 215-32. (6) Krewski, D. et al. Human and Ecological Risk Assessment, in press. (7) Slovic, F? et al. Risk Anal. 1995, in press. (8) Whittemore, A. S. Risk Anal. 1983,3(1),23-34. (9) Slovic, I? In Acceptable Evidence: Science and Values in Risk Management; Mayo, D. G.; Hollander, R. D., Eds.; Oxford Universitv Press: New York, 1991; DD. 48-65. Wddavsh, A. Searching for Safety-’social Philosophy and Policy Center and Transaction Publishers: New Brunswick, NJ, 1988. Ames, B. N.; Gold, L. S. Proc. Natl. Acad. Sci. USA 1990, 87(19), 7777-81. Sielken, R. L. Environ. Sci. Technol. 1987,21(11), 103339. Freedman, D. A.; Zeisel, H. Statistical Science 1988,3(1), 3-56. Bailar, S. C. I11 et al. Risk Anal. 1988,8(4), 485-97. Sielken, R. L.; Stevenson, D. E. Regul. Toxicol. Pharmacol. 1994, 19(1), 106-14. Abelson, I! H. Science 1994,265(5178), 1507. Tobacco:A Major International Health Hazard Zaridze, D.; Peto, R., Eds.; IARC ScientificPublications No. 74; International Agency for Research on Cancer: Lyon, 1986. Siemiatycki, J. et al. Int. J. Epidemiol. 1995, in press. Shields, I! G.; Harris, C. C. /.Am. Med. Assoc. 1991,266(5), 681-87. Mechanisms of Carcinogenesisin Risk IdentiJcation;Vainio, H.; Magee, I?; Macgregor, D.; McMichael, A. J., Eds.; IARC Scientific Publication No. 116; International Agency for Research on Cancer: Lyon, 1992. Cavenee, W. K.; White, R. L. Sci. Am. 1995,272(3), 7279. Bus, J. S.; Gibson, J. E. In Patty’s Industrial Hygiene and Toxicology, 2nd ed.; 3B Biological Responses, 1985;Vol. 111, pp. 143-74. Koshland, D. Science 1994,266(5193), 1925. Armitage, P: Environ. Health Perspect. 1985, 63, 195201. Moolgavkar, S. H.; Luebeck, G. RiskAnal. 1990,10(2),32341. Crump, K. S. et al. Cancer Res. 1976, 36(9), 2973-79. Hoel, D. G. Fed. Proc. 1980, 39(1),67-79. Clayson, D. B.; Krewski, D.; Munro, I. C. Regul. Toxicol. Pharmacol. 1983,3, 329-48. Staffa, J. A.; Mehlman, M. A. Innovations in Cancer Risk Assessment (ED,, Study); Pathotox Publishers: Park Forest South, IL, 1979. Brown, K. G.; Hoel, D. G. Fundam. Appl. Toxicol. 1983, 3(5), 470-77.
(31) Gaylor, D. W.; Frith, C. H.; Greenman, D. L. J. Environ. Pathol. Toxicol. Oncol. 1985,6(1), 127-36. (32) Peto, R. et al. Cancer Res. 1991,51(23part 21, 6415-51. (33) Lubin, J. H. et al. Radon and Lung Cancer Risk: A Joint Analysis of 11 Underground Miners Studies; National Cancer Institute: Washington, DC, 1994. (34) Lubin, J. H. Am. J. Epidemiol. 1994, 140(8),323-32. (35) IARC Study Group on Cancer Risk among Nuclear Workers. Lancet 1994,344(8929), 1039-43. (36) Lutz, W. K. Carcinogenesis (London) 1990, 11(8),124347. (37) Beland, EA.; Poirier M. C. Environ. Health Perspect. 1993, 99, 5-10. (38) Poirier, M. C.; Beland, E A. Chem. Res. Toxicol. 1992,5(6), 749-55. (39) Krewski, D. et al. Biometrics 1993, 49(2), 499-510. (40) Andersen, M. E. et al. Toxicol.App1.Pharmacol. 1987,87(2), 185-205. (41) Krewski, D. et al. Environ. Health Perspect. 1994, 102(1 SUPPI.l l ) ,37-50. (42) Ames, B. N.; Gold L. S. Science 1990,249(4972),970-71. (43) Huff, J. et al. Environ. Health Perspect. 1991,93,247-70. (44) Krewski, D. et al. RiskAnal. 1993, 13(4),463-78. (45) Apostoulu, A. Regul. Toxicol.Pharmacol. 1990,11(1),6880.
(46) Gaylor, D. W. Regul. Toxicol. Pharmacol. 1989,9(1), 1-8. (47) Committee on Risk Assessment Methodology. In Issues in Risk Assessment; National Research Council; National Academy Press: Washington, DC, 1993; pp. 1-183. (48) U.S. Environmental Protection Agency. Fed. Regist. 1986, 51(185),33992-34003. (49) Health and Welfare Canada. CarcinogenEvaluation;Health and Welfare Canada: Ottawa, 1992. (50) Kelly, K. E. Presented at the 84th Annual Meeting of the Air and Waste Management Association,Vancouver, June 1991; paper 91-175.4. (51) Sadouitz, M.; Graham, J. D. “Levels of Risk Allowed by U.S. Regulatory Standards”; Report prepared for Health Canada, Ottawa, 1994. (52) Rulis, A. M. Regul. Toxicol. Pharmacol. 1986, 7(2), 16068. (53) Dawson, M., Councilor of the Kwanlin Dun First Nation, Whitehorse,Yukon, personal communication, 1993. (54) Crump, K. S. J. Environ. Pathol. Toxicol. Oncol. 1984,5(4/ 51, 339-48. (55) Krewski, D.; Gaylor, D. W.; Szyszkowicz,M. Environ. Health Perspect. 1991, 90, 279-85. (56) Gaylor, D. W. J. Toxicol. Environ. Health 1983,11(3),32936. (57) Health Canada. Human Health Risk Assessment for Priority Substances;Canadian Environmental Protection Act; Canada Communication Group: Ottawa, 1994; En-40215/41E. (58) U.S. Environmental Protection Agency. Integrated Risk Information System [database],1993. (59) Health Effects Assessment Summary Tables. Annual FY 1992; U.S. Environmental Protection Agency; Office of Emergency and Remedial Response: Washington, DC, 1992. (60) Hazard Assessment for Polychlorinated Dibenzo-pdioxins; U.S. Environmental Protection Agency; Office for Health and Environmental Assessment:Washington, DC, 1985; EPA/600/8-84/014E (61) Flamm, W. G. et al. In De Minimis Risk; Whipple, C., Ed.; Plenum Press: NewYork, 1987; pp. 87-92. (62) Krewski, D.; Szyzskowicz,M.; Rosenkranz, H. Regul. Toxicol. Pharmacol. 1990,12(1),13-29. (63) Gold, L. S. et al. Enuiron. Health Perspect. 1984,58,9-319. (64) Gold, L. S. et al. Environ. Health Perspect. 1993,100, 65168. (65) Health Assessment Document for 2,3,7,8-Tetrachlorodibenzo-p-dioxin and Related Compounds, external review draft; U.S. Environmental Protection Agency; Office of Research and Development; Office of Health and Environmental Assessment:Washington, DC, 1994; EPA/ 600/BP-92/001a, OOlb, 001c. (66) Armstrong, V C.; Newhook, R. C. Regul. Toxicol. Pharmacol. 1992,15(2), 111-21. a 70-year lifetime, is shown. TCDD is included because its potency value far exceeds all other carcinogensby a large margin, even though it is not believed to be a DNA-reactivecarcinogen.
VOL. 29, NO. 8, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 7 5 A
