Toxicology Letters, 64/65 (1992) 435-492 0 1992 Elsevier Science Publishers B.V., All rights reserved 03784274/92/$5.99
485
Use of biomarkers in epidemiology: quantitative aspects L. Ehrenberg and M. Tarnqvist Department ofRadiobiology, Stockholm University Stockholm (Sweden) Key word: Adducts; Biomarkers; Risk estimation; Multiplicative model; Epidemiology
SUMMARY Cancer initiators (mutagens) present, due to the absence of definable no effect threshold, a special problem in toxicology, requiring a high sensitivity of detection methods. Disease epidemiology aiming at identification of carcinogens and quantification of associated risks has a low resolving power, the detectable incidence or mortality increments being often orders of magnitude larger than those which are of public concern. Other drawbacks of disease epidemiology is the long latency times and the influence of confounders. The use of genetic endpoints as biomarkers suffers from low cause specificity, although this drawback seems to be overcome, partly at least, by emerging methods for determination of mutation spectra at the DNA level. Proximal cancer initiators/mutagens are electrophilic compounds or metabolites that can react with nucleophilic atoms in nucleic acids and proteins. These reactions lead to ‘adducts’ that can be identified and quantified, e.g. in lymphocytes and erythrocytes in blood samples. The shift from biological observations to chemical analysis permits sufficient sensitivity, and measurement can be done shortly after onset of exposure. The well-defined life span of the adducts to hemoglobin (Hb) offer possibilities of dose calculation and risk estimation. For these reasons the measurement of adducts to Hb and DNA constitutes a powerful epidemiological tool, applications of which has been initiated in work environments and the general environment and also in the search for a priori unknown carcinogens.
INTRODUCTION
In the current development of methods for measurement of adducts and other biomarkers, the techniques have primarily been applied to heavily exposed populations which, besides animal experiments, offer important study material. The demonstration of correlations with disease incidence constitutes a kind of validation of the methodology, showing its use as an Correspondence to: L. Ehrenberg, Department of Radiobiology, Stockholm University, S-106
91 Stockholm, Sweden.
466
epidemiological tool, with an ability to contribute to the identification of etiological factors and to shed light on mechanisms. By and large this approach has remained qualitative, even when the adduct levels are quantified. The original suggestion of using adduct levels for sensitive monitoring of in vivo doses upon which risk estimation could be based [ll has so far rarely been followed. It appears that this quantitative aspect, risk being understood as the (average) probability of disease in exposed populations, has to be integrated into the further development of methods towards higher sensitivity and their application in epidemiological investigations. This necessity of integrating the quantitative aspect also has a bearing on adducts (and their sources> incidentally.found. The quantitative aspect is required for decisions whether the respective risks are of such concern that preventive measures are called for. It is also required for the clarification of the causes of background carcinogenesis. RISK MODEL
Classical cancer epidemiology has to a large extent been dealing with specific tumour types, with a more or less well-founded view of environmental carcinogens being site-specific. Oncologists often express the opinion that cancer is a collective term for a great number of diseases. Considering the present, albeit incomplete, knowledge on the mechanisms of tumour development, we should, in principle, count two types of environmental cancer risk factors: - Mutagens, i.e. compounds that are, or are metabolized to, electrophilic reagents [Zl, which through increase of the frequency of initiations lead to a general increase of the incidence of cancer; - Agents of various kinds (tissue-damaging factors, viruses, inhaled particles, irritants, etc.) that have a de-restraining effect, leading to release and growth of cancer cells; these agents may be referred to as promoters. Also modifying factors are perceived as carcinogens. Against this background a risk model will be presented and illustrated, mainly by data from studies of doses from ethene and its reactive metabolite, ethylene oxide. There is no absolute difference in action pattern of these types of factors. Initiators may show certain site specificities through dose gradients in the body, and certain compounds (e.g., benzo[alpyrene and many other PAHs) may be “complete” carcinogens, exhibiting actions of both types. This concerns also certain mixed exposures such as tobacco smoking. It is only the first type of risk factors (mutagens) that are generally measurable through adducts to DNA or proteins. With regard to cancer risks, on the other hand, the mutagens may seem more important in view of
the fact that dose responses for initiators should be considered - and probably are - linear, whereas dose-response relationships for the promoter action of chemicals may be described by S-shaped curves (such as the cumulative normal distribution) 131.At low exposure levels the latter may exhibit a no-effect threshold, unless their effects are added to those of ongoing related processes. At low exposure levels, chemical initiators are therefore expected to provoke an increased cancer risk, P,,(D), that is proportional to the dose D and to the background promotive situation which, including modifying factors, may be denoted PO&,as in Eqn. 1,
where a is the background initiation frequency and b is a proportionality parameter [31. This expression is compatible with the multiplicative model (Eqn. 2) lately found to be applicable to radiogenic cancer in experimental animals 141and humans 61, and which can also be fitted to incidences of ethylene oxide-induced tumours in rodents (Ehrenberg et al., to be published): Pcan-j(D)
=
(1 + p)
Em-j
(2)
where P&n-j is the background cancer incidence over sites j and l3 is a parameter.
METHODS
FOR DOSE MONITORING
AND RISK ESTIMATION
It was indicated early on that the relative mutagenic potencies, Qi, of chemicals i, with gamma-radiation as reference standard, assumed approximately the same value in widely different organisms, and it appeared possible to estimate cancer risk increments from values for the radiationdose equivalents (i-ad-equivalents) of target doses, once these doses could be measured. The chemical dose D is preferentially defined by the time-integral of concentration, with the dimension concentration x time, e.g. millimolar-hour (mMh) [61. For ethylene oxide 1 mMh has approximately the same mutagenic action as 80 rad (0.8 Gy) of gamma-radiation. Applying the multiplicative model (Eqn. 2) for cancer risk at low doses and low dose rates available data are compatible with B=O.l% per rad of gamma-radiation or per rad-equivalent of a chemical mutagen administered to a population of western age distribution (Ehrenberg et al., to be published). The National Research Council 151calculates here with a circa 4 times greater risk, evidently in order to be “on the safe side”.
488
In vivo doses would be most naturally measured through the accumulated levels of adducts to DNA, i.e. the essential target structure for genotoxic action. Partly because of the variable rates of DNA repair and cell renewal, the kinetics of which are still insufficiently known, hemoglobin (Hb) was chosen, because of its well-defined life span (circa 4 months in healthy persons), as a surrogate dose monitor, and information on dose gradients has so far been obtained from short-term studies of animal models. Following acute exposure at dose D for a reactive compound or metabolite an incremental adduct level is related to dose through
WI WI
-_=k.D
(3)
where RY is the adduct, Y the nucleophile (Hb) and k the second-order rate constant for the formation of the adduct. The steady-state level reached after long exposure corresponds to the cumulative level in 9 weeks (one-half of the erythrocyte life span), and from this value the annual dose can be calculated provided the exposure during the last few months may be considered representative of the whole year. In discussions of radiological protection, the ICRP [71had set the dose limit to members of the public with regard to a risk increment of cancer death in the range of 10-6-10” per year being acceptable. This is 0.03-0.3% of the background annual risk of cancer death (circa 300.lo? and would be reached by a lifelong (70 years) exposure to about 10 G-50) mrad/year. The ability to detect doses from single genotoxic chemicals was preliminarily set at this limit in the development of a sensitive procedure, theN-alkyl Edman method, for the mass-spectrometric measurement of adducts to the N-termini (valines) of Hb 181.For ethylene oxide and many alkylating agents with similar reaction patterns, the limit corresponds to a detection level at about 1 pmol valine-N adducts per g globin, and this goal was achieved.
ESTIMATED RISKS TO POPULATIONS AND INDIVIDUALS
Dosimetry in the sense of Eqn. 3 above by means of adducts to N-termini and other nucleophilic sites of Hb has so far been carried out with a series of alkylating agents (epoxides, methanesulfonates, alkyl halides) and their precursors (e.g., alkenes). Following reduction of Schiff bases to N-alkylvalines, a number of aldehydes have also been determined [9]. Ethylene oxide and its precursor alkene, ethene, have played a role as a model in the development of methods. Estimated risks associated with various exposure situations are summarized in Table I.
439
TABLE I RISK ESTIMATION OF ETHYLENE OXIDE AND ETHENE Exposure
Steady-state level bmoVg)
Rad-equiv. annual dose
Associated lifetime risk of cancer death from 1 year’s exposure; x lo5
Ethylene oxide, 1 ppm, 40 h/week
2400
22
400
Ethene, 1 ppm, 40 h/week
120
1
20
Ethene from smoking 10 cig./day
85
0.8
16
Background
20 (8-25)
0.18
3.6
The relationships between in vivo dose and exposure level are uncertain mainly because of difficulties of correctly assessing representative average exposure levels in work environments. As a matter of fact adduct levels give mostly the best measure of the average exposure level. The dose from ethene in work environments has an additional uncertainty due to difficulties of correctly assessing the fraction of inhaled ethene that is converted metabolically to ethylene oxide. The risks are computed under the assumption that for this compound the dose is distributed evenly in the body, as has been found in rodent models [IO]. The estimated risks are higher by one order of magnitude than the values computed from the unit risk factors of the U.S. EPA [Ill. This discrepancy may be explained by too short follow-up periods in epidemiological studies, with preponderant observation of short-latency leukemias, and by not calculating with lifetime dose in extrapolations from animal-test data WI. Ethylene oxide resembles in this respect the A-bomb radiations in Hiroshima and Nagasaki, where leukemias were seen as the predominant malignancies in the first 15-20 years after the exposure. Like several other adducts, those from ethene/ethylene oxide exhibit a background level, amounting to some 20 pmol/g globin. By comparison with measurements of expired ethene, endogenously produced ethene - according to animal studies originating mainly from the intestinal flora and dietary factors - could be shown to be the main source of the background [131. Individuals vary in sensitivity to carcinogens, and therefore the risks discussed above are mean values for studied populations. The variation is partly heritable, partly acquired, particularly through enzyme inductions. The risk, i.e. the probability that an exposure to a procarcinogen will lead to cancer, is affected by the activity of bioactivating enzymes (e.g., cytochromes P450) and, more importantly, by detoxifying enzymes (e.g., glutathione transferases and epoxide hydrolases), the net effect of which is measured by
490
adduct levels. In the chain of events leading to tumour growth the risk is further influenced by the activities of enzymes for repair of DNA damage and by the status of determinants of growth propensity, particularly inherited active oncogenes. The development of molecular methods, along the lines discussed at the VIth IUTOX Congress, will contribute to the understanding of the causes of human cancer and is expected to make it possible to predict individuals’ sensitivity - irrespective of whether such knowledge is desired. For the clarification of the causes of background cancer, and for the measurement of small in vivo doses above a large noise subjected to heritable variation, twin studies have been shown to be a useful tool 1141. DISCUSSION: A LOOK AHEAD
Doses of electrophilic compounds/metabolites may with sufficient sensitivity be measured through their adducts to hemoglobin and this will certainly become possible also for DNA adducts, with better knowledge of the kinetics for their disappearance, and with improved methods for their chemical identification. Biochemical methods for identification of sensitive individuals are rapidly developing. In order to increase the reliability and acceptance of cancer risk estimates of mutagens/initiators based on biomonitoring signals such as adduct levels, a few problems have to be solved: (a) Clarification of the true relationship between dose and risk at very low doses and dose rates. (b) Decisive testing of the hypothesis that chemical initiators with a dose contribution all over the body lead to life-long risk increments similar to those demonstrated for low-LET radiation 151.(The expectation that the model electrophile discussed here, ethylene oxide, will give rise to tumours at most sites with a background incidence is, however,.already supported by results of animal experiments; WI and to be published). From the point of view of effective cancer prevention it is important to note that, except the consequences of tobacco use, the contribution to the total number of cancer cases in western countries due to human carcinogens so far identified is small compared to the background cancer incidence in knowingly unexposed groups. It is of theoretical and practical interest to know to what extent initiators of exogenous or endogenous origin participate in the development of the background cancer, e.g. among the dietary and other life-style factors which have been suggested to be a main cause 1151, and it would be of importance to find ways to subject this incidence to preventive measures. It is of interest in this context that background ethene, originating partly from dietary compounds and gut bacteria, is judged, by
491
the rad-equivalence approach, to act as an initiator in about 1% of the cancer cases among non-smokers in Sweden. To solve the problem of the background carcinogenesis methods should be developed for the identification and quantification of a priori unknown adducts in tissue specimens from knowingly unexposed individuals, and these methods should be applied in conjunction with biochemical criteria, e.g. for deficiencies in detoxification and repair functions [161. A method under development with a great potential for giving essential contributions to the solution of several of the problems of carcinogenesis is the “mutational spectrometry” by a combination of PCR and denaturing gradient gel electrophoresis [171. Due to the uniqueness of the spectra produced by different mutagens it seems possible to clarify the relative role of electrophiles identified through their adducts, and the method seems to be sensitive enough to give important data on dose-response relationships at low doses. ACKNOWLEDGEMENT
The reviewed work was supported financially by the Swedish Environmental Protection Agency and the U.S. Department of Energy (DE-FGOZ89ER60784). REFERENCES Ehrenberg, L. (1974) Genetic toxicity of environmental chemicals. Acta Biol. Iugosl. Ser. F Genetika 6,367-398. Miller, J.A. and Miller, E.C. (1977) Ultimate chemical carcinogens as reactive mutagenic electrophiles. In: H. Hiatt, J.D. Watson and J.A. Winsten (Eds.), Origins of Human Cancer, Book B. Mechanisms of Carcinogenesis. Cold Spring Harbor Laboratory, pp. 665-627. Ehrenberg, L. and Scalia-Tomba, G. (1991) Mathematical models for the initiating and promotive action of carcinogens. In: L. Hothorn (Ed.), Statistical Methods in Toxicology. Springer, Berlin. pp. 65-78. Storer, J.B., Mitchell, T.J. and Fry, R.J.M. (1988) Extrapolation of the relative risk of radiogenic neoplasms across mouse strains and to man. Radiat. Risk. 114.331353. National Research Council (1999) BEIR V Report. Health Effects of Exposure to Low Levels of Ionizing Radiation. National Academy Press, Washington, D.C. Ehrenberg, L., Moustacchi, E., Osterman-Golkar, S. and Ekman, G. (1983) Dosimetry of genotoxic agents and dose-response relationships of their effects. Mutat. Res. 123,121182. ICRP (1977) Publ. No. 26. Recommendations of the International Commission on Radiological Protection. Pergamon Press, Oxford. Tornqvist, M., Mowrer, J., Jensen, S. and Ehrenberg, L. (1986) Monitoring ofenvironmental cancer initiators through hemoglobin adducts by a modified Edman degradation method, Anal. B&hem. 154,255-266. Kautiainen, A. Tiirnqvist, M., Svensson, K. and Osterman-Golkar, S. (1989) Adducts of malonaldehyde and a few other aldehydes to hemoglobin. Carcinogenesis 10,2123-2130.
IQ Segerback, 11.(1983) Alkylation of DNA and hemoglobin in the mouse following exposure to ethene and ethene oxide. Chem.-Biol. Interact. 45,139-15X 11 Tornqvist, M. Segerbiick, D. and Ehrenberg, L. (1991) The ‘rad-equivalence approach’ for assessment and evaluation of cancer risks, exemplified by studios of ethylene oxide and ethene. In: R,C. Garner, P.B. Farmer, G.T. Steel and AS. Wright (Eds.), Human Carcinogen Exposure: Biomonitoring and Risk Assessment. Oxford University Press, Oxford, pp. 141-155. 12 TGrnqvist, M. and Ehrenberg, L. On the cancer risk ofurban air pollution. Environ. Health Perspect. (Submitted~ 13 Filser, J., Denk, B., Tiirnqvist, M,, Kessler, W. and Ehrenberg, L. (lQQ2) Pharmacokinetics of ethylene in man: Body burden with ethylene oxide and hydroxyethylation of hemoglobin due to endogenous and environmen~l ethylene. Arch. Toxicol. 66,157-163. 14 Tljrnqvist, M., Svartengren, M. and Ericsson, C.H. (1992) Methylations of hemoglobin from twins discordant for cigarette smoking: hereditary and tobacco related factors. Chem.-Bio~. Interact. 82,91-98. 15 Higginson, J, and Muir, C.S. (19’79) Environmental earcinogenesis: misconceptions and limitations to cancer control. J. Natl. Cancer Inst. 63,1291-1298. 16 Herrlich, P.A. (1992) Induction of gene expression by radiation. In: W.C. Dewey et af. (Eds.1 Radiation Research. Vol. 2. Proceedings, fin press). 1’7 Thilly, W.G. (1991) Mutational spectrometry: opportunity and limitations in human risk assessment. In: R.C. Garner, P.B. Farmer, G.T.Steel and A.S. Wright (Eds.) Human Carcinogen Exposure: Biomoni~ring and Risk Assessment. Oxford University Press, Oxford, pp. 127-133.
485
Use of biomarkers in epidemiology: quantitative aspects L. Ehrenberg and M. Tarnqvist Department ofRadiobiology, Stockholm University Stockholm (Sweden) Key word: Adducts; Biomarkers; Risk estimation; Multiplicative model; Epidemiology
SUMMARY Cancer initiators (mutagens) present, due to the absence of definable no effect threshold, a special problem in toxicology, requiring a high sensitivity of detection methods. Disease epidemiology aiming at identification of carcinogens and quantification of associated risks has a low resolving power, the detectable incidence or mortality increments being often orders of magnitude larger than those which are of public concern. Other drawbacks of disease epidemiology is the long latency times and the influence of confounders. The use of genetic endpoints as biomarkers suffers from low cause specificity, although this drawback seems to be overcome, partly at least, by emerging methods for determination of mutation spectra at the DNA level. Proximal cancer initiators/mutagens are electrophilic compounds or metabolites that can react with nucleophilic atoms in nucleic acids and proteins. These reactions lead to ‘adducts’ that can be identified and quantified, e.g. in lymphocytes and erythrocytes in blood samples. The shift from biological observations to chemical analysis permits sufficient sensitivity, and measurement can be done shortly after onset of exposure. The well-defined life span of the adducts to hemoglobin (Hb) offer possibilities of dose calculation and risk estimation. For these reasons the measurement of adducts to Hb and DNA constitutes a powerful epidemiological tool, applications of which has been initiated in work environments and the general environment and also in the search for a priori unknown carcinogens.
INTRODUCTION
In the current development of methods for measurement of adducts and other biomarkers, the techniques have primarily been applied to heavily exposed populations which, besides animal experiments, offer important study material. The demonstration of correlations with disease incidence constitutes a kind of validation of the methodology, showing its use as an Correspondence to: L. Ehrenberg, Department of Radiobiology, Stockholm University, S-106
91 Stockholm, Sweden.
466
epidemiological tool, with an ability to contribute to the identification of etiological factors and to shed light on mechanisms. By and large this approach has remained qualitative, even when the adduct levels are quantified. The original suggestion of using adduct levels for sensitive monitoring of in vivo doses upon which risk estimation could be based [ll has so far rarely been followed. It appears that this quantitative aspect, risk being understood as the (average) probability of disease in exposed populations, has to be integrated into the further development of methods towards higher sensitivity and their application in epidemiological investigations. This necessity of integrating the quantitative aspect also has a bearing on adducts (and their sources> incidentally.found. The quantitative aspect is required for decisions whether the respective risks are of such concern that preventive measures are called for. It is also required for the clarification of the causes of background carcinogenesis. RISK MODEL
Classical cancer epidemiology has to a large extent been dealing with specific tumour types, with a more or less well-founded view of environmental carcinogens being site-specific. Oncologists often express the opinion that cancer is a collective term for a great number of diseases. Considering the present, albeit incomplete, knowledge on the mechanisms of tumour development, we should, in principle, count two types of environmental cancer risk factors: - Mutagens, i.e. compounds that are, or are metabolized to, electrophilic reagents [Zl, which through increase of the frequency of initiations lead to a general increase of the incidence of cancer; - Agents of various kinds (tissue-damaging factors, viruses, inhaled particles, irritants, etc.) that have a de-restraining effect, leading to release and growth of cancer cells; these agents may be referred to as promoters. Also modifying factors are perceived as carcinogens. Against this background a risk model will be presented and illustrated, mainly by data from studies of doses from ethene and its reactive metabolite, ethylene oxide. There is no absolute difference in action pattern of these types of factors. Initiators may show certain site specificities through dose gradients in the body, and certain compounds (e.g., benzo[alpyrene and many other PAHs) may be “complete” carcinogens, exhibiting actions of both types. This concerns also certain mixed exposures such as tobacco smoking. It is only the first type of risk factors (mutagens) that are generally measurable through adducts to DNA or proteins. With regard to cancer risks, on the other hand, the mutagens may seem more important in view of
the fact that dose responses for initiators should be considered - and probably are - linear, whereas dose-response relationships for the promoter action of chemicals may be described by S-shaped curves (such as the cumulative normal distribution) 131.At low exposure levels the latter may exhibit a no-effect threshold, unless their effects are added to those of ongoing related processes. At low exposure levels, chemical initiators are therefore expected to provoke an increased cancer risk, P,,(D), that is proportional to the dose D and to the background promotive situation which, including modifying factors, may be denoted PO&,as in Eqn. 1,
where a is the background initiation frequency and b is a proportionality parameter [31. This expression is compatible with the multiplicative model (Eqn. 2) lately found to be applicable to radiogenic cancer in experimental animals 141and humans 61, and which can also be fitted to incidences of ethylene oxide-induced tumours in rodents (Ehrenberg et al., to be published): Pcan-j(D)
=
(1 + p)
Em-j
(2)
where P&n-j is the background cancer incidence over sites j and l3 is a parameter.
METHODS
FOR DOSE MONITORING
AND RISK ESTIMATION
It was indicated early on that the relative mutagenic potencies, Qi, of chemicals i, with gamma-radiation as reference standard, assumed approximately the same value in widely different organisms, and it appeared possible to estimate cancer risk increments from values for the radiationdose equivalents (i-ad-equivalents) of target doses, once these doses could be measured. The chemical dose D is preferentially defined by the time-integral of concentration, with the dimension concentration x time, e.g. millimolar-hour (mMh) [61. For ethylene oxide 1 mMh has approximately the same mutagenic action as 80 rad (0.8 Gy) of gamma-radiation. Applying the multiplicative model (Eqn. 2) for cancer risk at low doses and low dose rates available data are compatible with B=O.l% per rad of gamma-radiation or per rad-equivalent of a chemical mutagen administered to a population of western age distribution (Ehrenberg et al., to be published). The National Research Council 151calculates here with a circa 4 times greater risk, evidently in order to be “on the safe side”.
488
In vivo doses would be most naturally measured through the accumulated levels of adducts to DNA, i.e. the essential target structure for genotoxic action. Partly because of the variable rates of DNA repair and cell renewal, the kinetics of which are still insufficiently known, hemoglobin (Hb) was chosen, because of its well-defined life span (circa 4 months in healthy persons), as a surrogate dose monitor, and information on dose gradients has so far been obtained from short-term studies of animal models. Following acute exposure at dose D for a reactive compound or metabolite an incremental adduct level is related to dose through
WI WI
-_=k.D
(3)
where RY is the adduct, Y the nucleophile (Hb) and k the second-order rate constant for the formation of the adduct. The steady-state level reached after long exposure corresponds to the cumulative level in 9 weeks (one-half of the erythrocyte life span), and from this value the annual dose can be calculated provided the exposure during the last few months may be considered representative of the whole year. In discussions of radiological protection, the ICRP [71had set the dose limit to members of the public with regard to a risk increment of cancer death in the range of 10-6-10” per year being acceptable. This is 0.03-0.3% of the background annual risk of cancer death (circa 300.lo? and would be reached by a lifelong (70 years) exposure to about 10 G-50) mrad/year. The ability to detect doses from single genotoxic chemicals was preliminarily set at this limit in the development of a sensitive procedure, theN-alkyl Edman method, for the mass-spectrometric measurement of adducts to the N-termini (valines) of Hb 181.For ethylene oxide and many alkylating agents with similar reaction patterns, the limit corresponds to a detection level at about 1 pmol valine-N adducts per g globin, and this goal was achieved.
ESTIMATED RISKS TO POPULATIONS AND INDIVIDUALS
Dosimetry in the sense of Eqn. 3 above by means of adducts to N-termini and other nucleophilic sites of Hb has so far been carried out with a series of alkylating agents (epoxides, methanesulfonates, alkyl halides) and their precursors (e.g., alkenes). Following reduction of Schiff bases to N-alkylvalines, a number of aldehydes have also been determined [9]. Ethylene oxide and its precursor alkene, ethene, have played a role as a model in the development of methods. Estimated risks associated with various exposure situations are summarized in Table I.
439
TABLE I RISK ESTIMATION OF ETHYLENE OXIDE AND ETHENE Exposure
Steady-state level bmoVg)
Rad-equiv. annual dose
Associated lifetime risk of cancer death from 1 year’s exposure; x lo5
Ethylene oxide, 1 ppm, 40 h/week
2400
22
400
Ethene, 1 ppm, 40 h/week
120
1
20
Ethene from smoking 10 cig./day
85
0.8
16
Background
20 (8-25)
0.18
3.6
The relationships between in vivo dose and exposure level are uncertain mainly because of difficulties of correctly assessing representative average exposure levels in work environments. As a matter of fact adduct levels give mostly the best measure of the average exposure level. The dose from ethene in work environments has an additional uncertainty due to difficulties of correctly assessing the fraction of inhaled ethene that is converted metabolically to ethylene oxide. The risks are computed under the assumption that for this compound the dose is distributed evenly in the body, as has been found in rodent models [IO]. The estimated risks are higher by one order of magnitude than the values computed from the unit risk factors of the U.S. EPA [Ill. This discrepancy may be explained by too short follow-up periods in epidemiological studies, with preponderant observation of short-latency leukemias, and by not calculating with lifetime dose in extrapolations from animal-test data WI. Ethylene oxide resembles in this respect the A-bomb radiations in Hiroshima and Nagasaki, where leukemias were seen as the predominant malignancies in the first 15-20 years after the exposure. Like several other adducts, those from ethene/ethylene oxide exhibit a background level, amounting to some 20 pmol/g globin. By comparison with measurements of expired ethene, endogenously produced ethene - according to animal studies originating mainly from the intestinal flora and dietary factors - could be shown to be the main source of the background [131. Individuals vary in sensitivity to carcinogens, and therefore the risks discussed above are mean values for studied populations. The variation is partly heritable, partly acquired, particularly through enzyme inductions. The risk, i.e. the probability that an exposure to a procarcinogen will lead to cancer, is affected by the activity of bioactivating enzymes (e.g., cytochromes P450) and, more importantly, by detoxifying enzymes (e.g., glutathione transferases and epoxide hydrolases), the net effect of which is measured by
490
adduct levels. In the chain of events leading to tumour growth the risk is further influenced by the activities of enzymes for repair of DNA damage and by the status of determinants of growth propensity, particularly inherited active oncogenes. The development of molecular methods, along the lines discussed at the VIth IUTOX Congress, will contribute to the understanding of the causes of human cancer and is expected to make it possible to predict individuals’ sensitivity - irrespective of whether such knowledge is desired. For the clarification of the causes of background cancer, and for the measurement of small in vivo doses above a large noise subjected to heritable variation, twin studies have been shown to be a useful tool 1141. DISCUSSION: A LOOK AHEAD
Doses of electrophilic compounds/metabolites may with sufficient sensitivity be measured through their adducts to hemoglobin and this will certainly become possible also for DNA adducts, with better knowledge of the kinetics for their disappearance, and with improved methods for their chemical identification. Biochemical methods for identification of sensitive individuals are rapidly developing. In order to increase the reliability and acceptance of cancer risk estimates of mutagens/initiators based on biomonitoring signals such as adduct levels, a few problems have to be solved: (a) Clarification of the true relationship between dose and risk at very low doses and dose rates. (b) Decisive testing of the hypothesis that chemical initiators with a dose contribution all over the body lead to life-long risk increments similar to those demonstrated for low-LET radiation 151.(The expectation that the model electrophile discussed here, ethylene oxide, will give rise to tumours at most sites with a background incidence is, however,.already supported by results of animal experiments; WI and to be published). From the point of view of effective cancer prevention it is important to note that, except the consequences of tobacco use, the contribution to the total number of cancer cases in western countries due to human carcinogens so far identified is small compared to the background cancer incidence in knowingly unexposed groups. It is of theoretical and practical interest to know to what extent initiators of exogenous or endogenous origin participate in the development of the background cancer, e.g. among the dietary and other life-style factors which have been suggested to be a main cause 1151, and it would be of importance to find ways to subject this incidence to preventive measures. It is of interest in this context that background ethene, originating partly from dietary compounds and gut bacteria, is judged, by
491
the rad-equivalence approach, to act as an initiator in about 1% of the cancer cases among non-smokers in Sweden. To solve the problem of the background carcinogenesis methods should be developed for the identification and quantification of a priori unknown adducts in tissue specimens from knowingly unexposed individuals, and these methods should be applied in conjunction with biochemical criteria, e.g. for deficiencies in detoxification and repair functions [161. A method under development with a great potential for giving essential contributions to the solution of several of the problems of carcinogenesis is the “mutational spectrometry” by a combination of PCR and denaturing gradient gel electrophoresis [171. Due to the uniqueness of the spectra produced by different mutagens it seems possible to clarify the relative role of electrophiles identified through their adducts, and the method seems to be sensitive enough to give important data on dose-response relationships at low doses. ACKNOWLEDGEMENT
The reviewed work was supported financially by the Swedish Environmental Protection Agency and the U.S. Department of Energy (DE-FGOZ89ER60784). REFERENCES Ehrenberg, L. (1974) Genetic toxicity of environmental chemicals. Acta Biol. Iugosl. Ser. F Genetika 6,367-398. Miller, J.A. and Miller, E.C. (1977) Ultimate chemical carcinogens as reactive mutagenic electrophiles. In: H. Hiatt, J.D. Watson and J.A. Winsten (Eds.), Origins of Human Cancer, Book B. Mechanisms of Carcinogenesis. Cold Spring Harbor Laboratory, pp. 665-627. Ehrenberg, L. and Scalia-Tomba, G. (1991) Mathematical models for the initiating and promotive action of carcinogens. In: L. Hothorn (Ed.), Statistical Methods in Toxicology. Springer, Berlin. pp. 65-78. Storer, J.B., Mitchell, T.J. and Fry, R.J.M. (1988) Extrapolation of the relative risk of radiogenic neoplasms across mouse strains and to man. Radiat. Risk. 114.331353. National Research Council (1999) BEIR V Report. Health Effects of Exposure to Low Levels of Ionizing Radiation. National Academy Press, Washington, D.C. Ehrenberg, L., Moustacchi, E., Osterman-Golkar, S. and Ekman, G. (1983) Dosimetry of genotoxic agents and dose-response relationships of their effects. Mutat. Res. 123,121182. ICRP (1977) Publ. No. 26. Recommendations of the International Commission on Radiological Protection. Pergamon Press, Oxford. Tornqvist, M., Mowrer, J., Jensen, S. and Ehrenberg, L. (1986) Monitoring ofenvironmental cancer initiators through hemoglobin adducts by a modified Edman degradation method, Anal. B&hem. 154,255-266. Kautiainen, A. Tiirnqvist, M., Svensson, K. and Osterman-Golkar, S. (1989) Adducts of malonaldehyde and a few other aldehydes to hemoglobin. Carcinogenesis 10,2123-2130.
IQ Segerback, 11.(1983) Alkylation of DNA and hemoglobin in the mouse following exposure to ethene and ethene oxide. Chem.-Biol. Interact. 45,139-15X 11 Tornqvist, M. Segerbiick, D. and Ehrenberg, L. (1991) The ‘rad-equivalence approach’ for assessment and evaluation of cancer risks, exemplified by studios of ethylene oxide and ethene. In: R,C. Garner, P.B. Farmer, G.T. Steel and AS. Wright (Eds.), Human Carcinogen Exposure: Biomonitoring and Risk Assessment. Oxford University Press, Oxford, pp. 141-155. 12 TGrnqvist, M. and Ehrenberg, L. On the cancer risk ofurban air pollution. Environ. Health Perspect. (Submitted~ 13 Filser, J., Denk, B., Tiirnqvist, M,, Kessler, W. and Ehrenberg, L. (lQQ2) Pharmacokinetics of ethylene in man: Body burden with ethylene oxide and hydroxyethylation of hemoglobin due to endogenous and environmen~l ethylene. Arch. Toxicol. 66,157-163. 14 Tljrnqvist, M., Svartengren, M. and Ericsson, C.H. (1992) Methylations of hemoglobin from twins discordant for cigarette smoking: hereditary and tobacco related factors. Chem.-Bio~. Interact. 82,91-98. 15 Higginson, J, and Muir, C.S. (19’79) Environmental earcinogenesis: misconceptions and limitations to cancer control. J. Natl. Cancer Inst. 63,1291-1298. 16 Herrlich, P.A. (1992) Induction of gene expression by radiation. In: W.C. Dewey et af. (Eds.1 Radiation Research. Vol. 2. Proceedings, fin press). 1’7 Thilly, W.G. (1991) Mutational spectrometry: opportunity and limitations in human risk assessment. In: R.C. Garner, P.B. Farmer, G.T.Steel and A.S. Wright (Eds.) Human Carcinogen Exposure: Biomoni~ring and Risk Assessment. Oxford University Press, Oxford, pp. 127-133.
