Is ambient ethene a cancer risk factor?

Ethene is, on a molar basis, a major urban air pollutant. It has been shown beyond doubt that a fraction of inhaled ethene is metabolized in mammals (including humans) via ethylene oxide, an electrophilic reagent that has been shown to be mutagenic and carcinogenic. To the extent that the linearity hypothesis for dose-response relationships at low levels is accepted, exposure to ethene is therefore expected to lead to a risk increment. In order to judge whether ethene as a single compound should be considered a risk factor, it has to be evaluated whether this risk increment is negligibly small or of concern to individuals or societies. The magnitude of the cancer risk from ethene cannot be inferred from animal experiments. Because of saturation of the metabolism of ethene, sufficient statistical power cannot be attained in long-term animal tests with about 100 animals per dose. By application of the radiation-dose equivalent of the unit of target dose of ethylene oxide and using the best (although still uncertain) value for the conversion factor (about 5%), exposure to 10 ppb ethene--a level occurring in urban areas--is expected to lead to a lifetime risk of cancer death amounting to approximately 70 per 100,000. According to a recent estimate the average exposure in Sweden to ethene is some six times lower. These figures are uncertain by a factor of at least three. They indicate ethene to be a risk factor of concern.


Introduction
The gas phase of automotive engine exhausts and consequently of polluted urban air contains a number of unsaturated hydrocarbons (i.e., alkenes, alkadienes). Particularly ethene occurs at relatively high concentrations (1), some 105 times higher than that of, for example, benzok [o]pyrene. Alkenes are known to be metabolized in vivo via epoxides, as was first shown for ethene (2)(3)(4) and later for butadiene (4). Because 1,2-epoxides are electrophilically reactive and have been shown to be mutagenic and carcinogenic (5), the parent alkenes also should be suspected of being potential environmental carcinogens. The alkenes ethene (6), propene (7), and 1,3-butadiene (8) have been subjected to long-term cancer tests with rodents, only the latter showing a positive result.
The reason why a carefully conducted cancer test with ethene in the Fischer 344 rat did not show any change of the tumor incidence may be attributed to saturation of the metabolism, limiting the ethylene oxide dose achievable to approximately one-fourth of the dose that would permit detection, at reasonable statistical power, of raised cancer incidence in a test with some 100 animals ( Figure 1) (9,10). The positive response to butadiene probably is because of partial for-mation also of diepoxybutane, the mutagenic and carcinogenic effectiveness of which is greater than that of the monooxiranes by some two orders of magnitude (5).
This paper summarizes efforts to elucidate the question: To what extent is ambient ethene to be considered a risk factor of concern? This approach thus illustrates the quantitative aspect of risk identification, a risk-management function that is qualitative in principle (11).
The previously mentioned inability of ethene to provoke a significant increase in cancer tests of conventional scope might at first sight be taken to indicate that the compound is innocuous. In view of the low resolving power of cancer tests, however, this is not necessarily true. Considering that the metabolite, ethylene oxide, is judged to be a probable human carcinogen (12), it has to be investigated whether the cancer risk associated with ambient concentrations of ethene, in urban environments mostly at parts per billion levels, is of nonnegligible magnitude. Because it has not been possible to extrapolate the risk of ethene from animal test data, other ways must be sought out. These must comprise in vivo dose measurement of ethylene oxide, the reactive metabolite, and establishment of the relationship between in vivo dose and risk. In the present study, the latter has been done by determination of the radiation-dose equivalent (rad-equivalent) of a unit of chemical dose, given as millimolarhour (mMhr), the dose defined as the integral over time of concentration (13). Besides development of a technique for in vivo dose monitoring, a number of ques-tions concerning person-weighted exposure doses, metabolism in animals and man, etc., had to be answered in order to judge whether ethene should be considered a risk factor. The rad-equivalent approach appears to be useful for the identification and estimation of risks from many environmental genotoxic chemicals for which disease-epidemiological or cancer-test data are unavailable (14). Monitoring of the in vivo dose of a reactive chemical or its metabolites is based on the measurement of reaction products with (adducts to) macromolecules in tissues. This association of risk with a chemical end point increases the sensitivity by several orders of magnitude as compared to observation of biological end points, a fact

Methods
It was early shown by Miller and Miller (15) that most genotoxic chemicals (i.e., cancer initiators and mutagens) are electrophilic reagents or are metabolized to such reagents. Electrophiles (e.g., alkylating agents such as epoxides) react with nucleophilic atoms in DNA, the target for biological effects. Reaction products with nucleophilic atoms (S, N, 0) in other macromolecules, especially certain amino acids in proteins, are parallelly formed. Determination of adducts to DNA and proteins offers a means of identifying genotoxic risk factors in vivo. Hemoglobin (Hb) has been found suitable as a monitor molecule because it is accessible in large amounts and because Hb adducts have a long, welldefined life-span and can be identified chemically by available methods. Tissue doses useful for risk estimation can be calculated from levels of Hb adducts (16). Hb adduct monitoring was shown to be useful for risk identification and risk estimation, but the lack of a sensitive, rapid, and reproducible analytical method limited applications. For this reason, a large effort was put into the development of a new method for Hb adduct analysis with a sensitivity sufficient for applications connected with urban air pollution.
The nitrogen in the N-terminal amino acids, valines, in the two a-and 5-chains in Hb, is a major reaction site for epoxides and other alkylating agents (16). This was one reason to try the Edman sequencing technique for isolation of Hb adducts. When this procedure was applied to [ 14C]ethylene oxide-alkylated globin, it was observed (17) that radioactive material could be extracted already from the weakly alkaline coupling medium in which the protein is treated with the Edman reagent. This work resulted in a procedure, the N-alkyl Edman method (18,19), in which globin samples are derivatized with pentafluorophenyl isothiocyanate in formamide. Alkylated N-terminal valines are then extracted as pentafluorophenylthiohydantoins, which, following purification steps, are analyzed with high sensitivity by gas chromatography-mass spectrometry with chemical ionization in the negative ion mode.

Results
Doses ofEthylene Oxide in Ethene-exposed Humans A problem of basic importance in risk assessment of ethene is the relationship between exposure dose of ethene and dose in tissues of the ultimate carcinogen, ethylene oxide. In an early phase of this work, this ratio was established in animal models.
Segerback (3) measured, by means of Hb adducts, doses in mice that were compatible with 8% of inhaled ethene being metabolically converted to ethylene oxide. In this work, radiolabeled ethene was used, and from the level of DNA adducts in various tissues, it was concluded that the dose was approximately the same in different parts of the body. This means that the measured blood dose is relevant to the target dose in various organs. T6rnqvist et al. (19) used the N-alkyl Edman method to determine in vivo doses of ultimate carcinogens from alkenes in hamsters and rats exposed to automotive engine exhausts. Through the determination of N-(2hydroxyethyl) valine (HOEtVal), the adduct formed in the reaction of ethylene oxide with N-terminal valine, it was found that 5 to 10% of inhaled ethene in the exhaust was metabolized to ethylene oxide. The uncertainty in the determination was mainly due to difficulties in correctly assessing time-weighted average exposure concentrations. The exposures were carried out with diesel exhaust, with and without particle filter, three and two doses, respectively; and with gasoline exhaust, with and without catalyst, at two doses. Adduct levels were shown to be linearly dependent on exposure level. When a catalytic converter was used, the adduct levels in animals exposed to gasoline exhaust were strongly reduced in agreement with the reduced ethene exposure.
Although these experiments showed approximately the same rate of conversion of inhaled ethene to ethylene oxide in three animal species and thus invites to extrapolation to other species, it was considered important to prove that ethene is metabolized in the same way in man.
The conversion of ethene to ethylene oxide in humans has been studied in cigarette smokers; a steady-state adduct increment has been shown amounting to about 8 pmole HOEtVal/g globin/smoked cigarette and day (20,21), a value confirmed by Bailey et al. (22) using a modification of the N-alkyl Edman method. A conclusion from these studies is that about 6% of about 0.25 mg ethene in the mainstream smoke from each cigarette is converted to ethylene oxide. Studies of persons with occupational exposure to ethene e.g., fruitstore workers (23) and plastics industry workers (Tornqvist et al., unpublished material), show values in the range of 1 to 10% conversion. This uncertainty is mainly due to difficulties of correctly assessing the exposure to ethene. For that reason, human inhalation studies with controlled exposure have been conducted with results indicating about 2% conversion but a three times longer half-life in vivo of ethylene, i.e., in support of the in vivo dose per inhaled milligram of ethene as concluded from the smoker study (manuscript in preparation). In a pharmacokinetic study in humans, Shen et al. (24) found about 4% of inhaled ethene to be metabolized in the body.
The previously mentioned judgments were based on dose monitoring in persons occupationally exposed to ethylene oxide, where the steady-state level of HOEtVal was found to be 2.4 nmoles/g globin during exposure to 1 ppm ethylene oxide, 40 hr/week (25).

Hemoglobin Adducts firom Ethene in Urban Air
The present version of the N-alkyl Edman method is sufficiently sensitive to measure adduct levels down to a few pmole per gram globin (i.e., the increments expected to be associated with chronic exposure to a few parts per billion of ethene). Such measurements, however, are obscured by the existence of a background level of HOEtVal in Hb. In addition, epidemiological studies of the variation of the adduct level have not yet been carried out on populations sufficiently differing in degree of urbanization (26,21).
The background varies in the range from 8 to 25 pmole HOEtVal/g globin and has been found in animal experiments to be partly caused by ethene formation from food and intestinal bacteria (27). In measurements of ethene exhaled by humans, it was confirmed that endogenously formed ethene is the main source of the background HOEtVal level (28), allowing for a small contribution (about 5 pmole/g globin) from urban air and/or passive smoking (29). It has also been shown that at these low levels artifact formation of HOEtVal during storage of Hb samples must be prevented (30).
Measurement of ethene at four sites in Stockholm 1987 (31) showed average levels around 25 ppb in areas with heavy traffic. As average exposure doses for urban areas or the whole country had not been determined at that time and HOEtVal measurements because of the background gave too uncertain information, the time-weighted average levels of ethene were estimated indirectly via carboxyhemoglobin increments in urban citizens (32,33). These deliberations were compatible with an aver-Environmental Health Perspectives age ethene exposure to the Swedish population in the range 10 to 20 ppb. Although this level may still occur in very urbanized areas, the average exposure level in Sweden was recently estimated to about 1.5 ppb (see below). An ethene concentration of 10 ppb, a reasonable value in urban areas, is expected to give an incremental adduct level of 4 pmole HOEtVal/g globin. Because of the variable background, an increment of this size could be well determined only in very large populations. In Table 1, the estimated average increments of HOEtVal from various sources of ethylene oxide and ethene have been compiled.

Discussion
In the rad-equivalence approach, the dose of acute yor X-radiation, which produces the same frequency of mutation (or other genotoxic changes) as 1 mMhr of a chemical under study is determined. This approach is based on the unproved but reasonable assumption that an initiated cell has the same chance of giving rise to a tumor, irrespectively of whether it was initiated by a mutagenic chemical or by low-linear energy transfer radiation, chosen as reference standard because it is the environmental factor that is best investigated with respect to doserisk relationships. A comparison of this kind (of a genotoxic chemical with ionizing radiation) is facilitated by the fact that at low doses the effects of both types of agents may be fitted to linear dose-response curves. Through this approach influences of promotive and cocarcinogenic conditions prevailing in human populations are implicitly estimated and allowed for (14).
In a broad range of test systems, 1 mMhr of ethylene oxide gives approximately the same response as 80 rad (0.8 Gy) of y-radiation (34). Data from the various studies of uptake and metabolism of ethene and ethylene oxide are further compatible with ethene giving an in vivo dose of ethylene oxide, which is about 5% of the dose received following direct inhalation of ethylene oxide. The dose received by personnel occupationally exposed to ethylene oxide has been estimated to correspond to 20 rad-equivalent/year ppm, with 1600 hr exposure per year (25).
From these data, the rad-equivalent of the annual dose of ethylene oxide from ethene at an ambient concentration of 10 ppb is calculated to be about 0.05 rad-equivalent per year.
The cancer risk associated with an exposure to 10 ppb ethene would thus be about one-half of the risk because of background radiation (about 0.1 rad/year). An analysis ofvarious data indicates that at low-dose rate the risk may be some four times lower (35) than has been estimated by the National Research Council (36). According to this, the lifetime cancer mortality risk at 10 ppb ethene would be approximately 7 x 10 Recent figures (35,37) indicate the average ethene level in Sweden to be 1.8 pg/m3 (approximately 1.5 ppb). This would mean an average lifetime risk in Sweden (8.4 million) of 1 x 10 , corresponding to some 13 cancer deaths and about twice as many cases of disease annually in the country (35). These figures for the consequences of today's ethene exposure are thus some five times lower than the ones estimated to be valid 10 years ago. They might, if ethene were the sole urban air pollutant because of the evaluation philosophy chosen, be considered close to the lower limit of unacceptability.
This identification of ethene as a risk factor calls for two kinds of action: A more accurate estimation of the magnitude of this risk, which, given the exposure level assumed here, has to be considered uncertain by at least a factor three; and measures to be taken, such as catalytic conversion in fuel burning, for reduction of this risk.