Priority Substances List - Statement of the Science Report for Ethylene Oxide

3.0 Assessment of "Toxic" under CEPA 1999

3.1 CEPA 1999 64(a): Environment

The environmental risk assessment of a PSL substance is based on the procedures outlined in Environment Canada (1997a). Analysis of exposure pathways and subsequent identification of sensitive receptors are the basis for selection of environmental assessment endpoints (e.g., adverse reproductive effects on sensitive fish species in a community). For each endpoint, a conservative Estimated Exposure Value (EEV) is selected and an Estimated No-Effects Value (ENEV) is determined by dividing a Critical Toxicity Value (CTV) by an application factor. A hyperconservative or conservative quotient (EEV/ENEV) is calculated for each of the assessment endpoints in order to determine whether there is potential ecological risk in Canada. If these quotients are less than one, it can be concluded that the substance poses no significant risk to the environment, and the risk assessment is completed. If, however, the quotient is greater than one for a particular assessment endpoint, then the risk assessment for that endpoint proceeds to an analysis where more realistic assumptions are used and the probability and magnitude of effects are considered. This latter approach involves a more thorough consideration of sources of variability and uncertainty in the risk analysis.

3.1.1 Assessment endpoints

All reported Canadian releases of ethylene oxide are to air, and pathways analysis indicates that following release to air, ethylene oxide is unlikely to partition to other compartments in significant amounts. Once in the atmosphere, ethylene oxide is not expected to contribute to ground-level ozone or climate change, nor will it deplete the ozone layer. Its atmospheric half-life may range from 38 to 382 days. Its high water solubility may encourage some washout via precipitation; however, available evidence indicates that this removal mechanism has minimal impact.

Although releases to water and soil are not common, it is understood that some releases to these media may occur in the event of a spill or similar release scenario. Persistence in these media is not expected, as ethylene oxide has a high Henry's law constant (12,2-19.9 Pa·m³/mol), and the experimental data indicate that volatilization will occur rapidly from water (t_1/2 ~1 hour). Although no information was available regarding concentrations of ethylene oxide in wastewater discharged from Canadian manufacturing and processing operations, releases from these sources are expected to be minimal, especially when one considers temperatures and retention times in wastewater treatment processes. Based on these considerations, aquatic concentrations are expected to be negligible; therefore, adverse impacts to naturally occuring aquatic organisms are also considered negligible.

Given that the primary medium of release for ethylene oxide is the atmosphere and that the chemical's properties cause it to remain in and react in that compartment, the assessment endpoint for determination of toxicity under CEPA 1999 Paragraph 64(a) will be for atmospherically exposed organisms. One of the more significant effects that has been observed following atmospheric exposure is the induction of genetic mutations in microorganisms, plants and animals. Other effects observed in laboratory animals include carcinogenicity, reduced kidney and adrenal weights, and increased incidence of inflammatory lesions in the lungs, nasal cavity, trachea and internal ear, as well as decreases in weight, changes in posture, demyelination of parts of nervous tissue, decreased sperm count and induction of other adverse reproductive effects. Although evidence is strong concerning ethylene oxide-induced genotoxic and carcinogenic effects (see Sections 2.4.4.1 and 2.4.4.2), the actual population-level impact to wildlife from these endpoints is not completely clear when one considers population resilience, dose-response and induction frequency. Among the observed effects, adverse impacts on reproduction are decidedly the endpoint that would have the greatest potential to adversely impact wildlife population levels. Other effects may occur at slightly lower concentrations.

3.1.2 Environmental risk characterization

3.1.2.1 Terrestrial biota

There are only a few ambient measurements of ethylene oxide in Canada. Some limited additional data were identified for the urban area of Los Angeles, California (Havlicek et al. 1992). The ambient air concentrations from Los Angeles are likely to be higher than would be expected to occur in most Canadian situations. Los Angeles is located in a geographic area (i.e., a basin) that can cause reduced air movement and contribute to higher pollutant levels in the air. For a hyperconservative scenario, it will be assumed that the concentrations of ethylene oxide in Los Angeles are comparable to maximum concentrations that may be found in Canada. The maximum mean 24-hour ambient air concentration detected in the Los Angeles urban area was found to be 956 µg/m³ (95% confidence limit [CL] = 0.75-5600 µg/m³; n = 6) in May 1990 and will be used as the EEV to represent a worst-case Canadian atmospheric concentration.

The EEV for ethylene oxide is therefore 956 µg/m³.

Toxicity data are very limited for all of the environmental compartments. The most sensitive terrestrial organisms were found to be laboratory test rodents, which will be considered as surrogates for Canadian wild rodent species. The CTV is derived from a reproductive study by Snellings et al. (1982b); the effects reported in this study were determined to represent the most significant ecological endpoint in terms of the potential to adversely impact natural wildlife populations. Snellings et al. (1982b) exposed 3-to 4-week-old Fischer 344 rats to ethylene oxide vapour at 10, 33 and 100 ppm (18,3, 60.4 and 183 mg/m³) for 6 hours per day, 5 days per week, then exposed only females at the same rate from day 0 to day 19 of gestation, 7 days per week. The authors reported that at the highest exposure concentration (183 mg/m³), there was a significant drop in the number of pups born per litter. There were also fewer implantation sites and fewer pups born per implantation site.

The CTV for terrestrial animals is 183 mg/m³ from the rat study of Snellings et al. (1982b). To derive the ENEV from this study, a suitable application factor is used. The magnitude of the application factor takes into consideration the fact that the CTV is based on a relatively small data set and was the maximum concentration tested in the study. In addition, the study was conducted in the laboratory (not the field), and no statistics were applied to determine whether 183 mg/m³ is truly the lowest adverse effect concentration, nor was the study designed to make this determination. For these reasons, and because effects less clearly related to population-level impacts (e.g., decreased weight) were observed at slightly lower levels, a relatively large application factor of 100 is applied to the CTV, as outlined in Environment Canada (1997a), to determine the ENEV.

The ENEV for terrestrial biota is therefore 1830 µg/m³.

The hyperconservative quotient is calculated by dividing the EEV of 956 µg/m³ by the ENEV of 1830 µg/m³.

Since the hyperconservative quotient is less than one, the risks posed by chronic exposure of terrestrial biota to ethylene oxide in the Canadian environment are expected to be minimal.

3.1.2.2 Discussion of uncertainty

There are a number of sources of uncertainty in this environmental risk assessment. The primary source is the lack of Canadian ambient environmental concentration data. Measured atmospheric concentrations reported in the assessment are from California and a small number of Canadian locations. The use of measured ambient air concentrations is preferred to the use of modelled results, and although the EEV chosen is more than an order of magnitude higher than concentrations predicted near Canadian sources based on modelling studies, it represents a conservative estimate of a potential concentration in the Canadian atmosphere.

No ambient measurements of ethylene oxide in water, soil, sediment or groundwater in Canada were located during the literature search. This is due to a number of factors, including the lack of monitoring programs designed to measure ethylene oxide as well as the physical/chemical properties of ethylene oxide, which govern its fate and behaviour, resulting in limited entry into or rapid removal from environmental compartments other than air.

An additional source of uncertainty is the limited number of toxicity data for species in all environmental compartments. Ideally, there should be enough toxicity data to represent a wide variety of species in each environmental compartment. For example, freshwater fish are represented by fathead minnow and goldfish, but no other species. Similarly, only one freshwater aquatic invertebrate and one marine invertebrate are represented in the toxicity database. However, based on the atmosphere being the predominant compartment of environmental release, and considering the environmental fate and behaviour of ethylene oxide, the limitations in the toxicity data were considered admissible.

Similar limitations are associated with the data available on effects on terrestrial organisms. In addition, although mutagenic effects have been observed in a variety of terrestrial plants and mammals, their population-level impacts are uncertain. To account for these uncertainties, a relatively large application factor was used in the risk analysis to derive an ENEV.

Despite a few data gaps regarding the environmental concentrations and effects of ethylene oxide, the data available at this time are considered adequate for drawing a conclusion on the environmental risk of ethylene oxide in Canada.

3.2 CEPA 1999 64(b): Environment upon which life depends

The theoretical atmospheric lifetime of ethylene oxide is long enough to allow a small percentage of the amount emitted to reach the stratosphere; however, ethylene oxide does not degrade to an active intermediate and therefore does not induce the depletion of the ozone layer. Its POCP is considered insignificant, and its GWP is considered minimal.

3.3 CEPA 1999 64(c): Human health

3.3.1 Estimated population exposure

Information on monitored levels of ethylene oxide in air, drinking water and foodstuffs in Canada is exceedingly limited, being restricted to detection in a few samples of ambient and indoor air in a small monitoring survey.

Although available data are limited, deterministic estimates of total daily intakes of ethylene oxide for the general population in Canada were developed - primarily to compare relative contributions from various media -on the basis of the limited monitoring data in ambient and indoor air (Conor Pacific Environmental, 1998), a limited survey of levels in foodstuffs in Denmark (Jensen, 1988) and concentrations in drinking water and air predicted by the ChemCAN 4 regional fugacity model, when advective input from bordering U.S. states was included (Health Canada, 1999a). Based on this approach, estimated intakes (expressed as µg/kg-bw per day) of ethylene oxide from food exceeded those from air; however, the extent of the uncertainties associated with the estimates, particularly in foodstuffs (i.e., based on measured levels in a limited number of food products consumed in other countries, and incorporating highly conservative and uncertain estimates for the consumption of selected food products [i.e., spices]), precludes any degree of confidence in these conclusions.

Owing to these limitations, the focus of the remainder of this section and the basis for risk characterization is exposure in air. This approach is supported on the basis that all releases from point sources controllable under CEPA are to air, that ethylene oxide is generally transferred to air following release to other media and that it is not expected to accumulate in sediment or soil or bioaccumulate, as a result of its high water solubility and vapour pressure.

The concentration of ethylene oxide predicted for ambient air (i.e., 6,2 x 10^-3 µg/m³) by ChemCAN fugacity modelling was considered the basis for the minimum estimate of exposure via inhalation. Censored mean concentrations of ethylene oxide in outdoor and indoor air (i.e., 0.34 µg/m³ and 0.17 µg/m³, respectively), derived from the multimedia exposure study, were considered to represent the maximum concentrations to which the general population is exposed daily indoors and outdoors, respectively. Upper-bounding estimates of exposure via inhalation for the general population in Canada were based upon the maximum concentrations of ethylene oxide in outdoor and indoor air (i.e., 4.9 µg/m³ and 4.0 µg/m³, respectively) reported from the multimedia exposure study (Conor Pacific Environmental, 1998). Mean concentrations in ambient air sampled in Los Angeles, California, ranged from 0.038 to 955,7 µg/m³ (Havlicek et al. 1992)

Exposure to ethylene oxide in ambient air may be substantially higher for populations residing in the vicinity of point sources. A concentration of 2 µg ethylene oxide/m³ was predicted for outdoor air in close proximity to hospitals in Canada (Environment Canada, 1999) and Florida (Tutt and Tilley, 1993). A concentration of 11 µg ethylene oxide/m³ was predicted for outdoor air in close proximity to a sterilization facility in Florida (Tutt and Tilley, 1993). A maximum 1-hour concentration of 20.1 µg ethylene oxide/m³ was predicted for outdoor air near a production facility for ethylene glycol in Alberta (Environment Canada, 1997b).

Limitations of the data preclude development of meaningful probabilistic estimates of exposure of the general population to ethylene oxide in air.

3.3.2 Hazard characterization

Owing to the physical/chemical properties of ethylene oxide, most toxicological studies have involved inhalation, which is the principal route of exposure of the general population in the vicinity of sources. There have been only a few investigations of potential health effects associated with the ingestion of this substance.

Pathways for the metabolism of ethylene oxide and subsequent elimination of its metabolites involve either hydrolysis or enzymatic conjugation with glutathione catalyzed by the GSTT1 enzyme. The parent compound is the putative toxin, interacting directly with DNA and proteins. Data from laboratory animals are consistent, in part, with glutathione conjugation being a detoxification pathway, with interspecies variations in toxicity correlating with greater specific activity of cytosolic GSTT1 in smaller species (i.e., mice versus rats). Available data indicate that the metabolism of ethylene oxide in humans and laboratory animals is qualitatively similar, although there may be quantitative variations, since conversion through hydrolysis appears to predominate in larger species (such as dogs). A genetic polymorphism in the expression of the GSTT1 enzyme in humans contributes to potential for considerable intraspecies (interindividual) variation in metabolism and, hence, response to ethylene oxide, which has been confirmed in in vitro studies, although the relative importance of this pathway in the detoxification of ethylene oxide in humans is unknown.

3.3.2.1 Carcinogenicity

Information relevant to assessment of the ca rcin ogen icit y of ethylene oxide has been derived from epidemiological studies of occupationally exposed workers, carcinogenesis bioassays in laboratory animals, as well as supporting data related to genotoxicity and metabolism.

While increases in mortality due to liver, colon, breast, bladder, kidney, esophageal, stomach, brain or pancreatic cancer have occasionally been reported in epidemiological studies of workers exposed to ethylene oxide, evidence is not consistent or convincing.

However, within the limits of sensitivity of identified studies, available epidemiological evidence for an association between exposure to ethylene oxide and lymphopoietic/hematopoietic cancer is suggestive, although inconclusive, based on consideration of traditional criteria for causality, as outlined below (e.g., consistency, specificity, dose-response relationship and biological plausibility).

Although not reported in all studies and generally based on small numbers of observed cases, increased risks of leukemia, all hematopoietic neoplasms (or non-Hodgkin's lymphoma in the same cohort), lymphopoietic/ hematopoietic cancers or lymphosarcoma/ reticulosarcoma have been reported for production and sterilization workers (Hogstedt, 1988), workers in plants producing sterilized medical supplies and spices (Steenland et al. 1991; Stayner et al. 1993; Wong and Trent, 1993) and those producing disposable medical equipment (Hagmar et al. 1995), respectively (Table 5). It is of interest that these excesses occurred in workers exposed primarily to ethylene oxide in the sterilization of medical supplies and equipment rather than in facilities associated with its production and/or use, where numerous other substances would have been present.

Risks for lymphopoietic/hematopoietic cancers among the various industrial cohorts have varied, although in general by less than 2-fold. For example, in meta-analyses, while the risk for leukemia was not significantly increased (sSMR = 1.06; 95% CI = 0.73-1.48 or 0.55-2.02, corrected for heterogeneity), the sSMR for non-Hodgkin's lymphoma was suggestively although not significantly increased (1.35; 95% CI = 0.93-1.90) (Shore et al. 1993). However, it should be noted that with the single exception of the investigation of workers at ethylene production plants in which there was no increase in hematopoietic cancers reported (Teta et al. 1993), length of follow-up was relatively short in the critical investigations, averaging 11.6 years and 16 years for the more reliable studies in which excesses were observed - namely, Hagmar et al. (1995) and Steenland et al. (1991). In the largest investigation (Steenland et al. 1991; Stayner et al. 1993), only 28% of workers had attained greater than 20 years since first exposure, and five of the seven men who died of leukemia did so within the most recent calendar period. Limited strength of the observed associations could be due, therefore, at least in part, to the limited period of follow-up.

In meta-analysis, although risks associated with the frequency or intensity of ethylene oxide exposure could be examined in only three studies, no trends were observed in either the individual or combined studies, although the positive trend by cumulative exposure in the largest investigation was noted (Shore et al. 1993). There were no trends in the individual or combined studies with respect to duration of exposure or latency. However, in the largest cohort examined (more than 18 000 workers who had been exposed primarily to ethylene oxide) with the most extensive characterization of individual exposure and the only investigation in which cumulative exposure was quantitatively estimated, regression analysis revealed a highly significant (p < 0.01) exposure-response relationship between cumulative exposure to ethylene oxide and mortality due to lymphocytic leukemia and non-Hodgkin's lymphoma combined (termed "lymphoid" neoplasms) (Steenland et al. 1991; Stayner et al. 1993). An association was also observed between cumulative exposure to ethylene oxide and mortality from all hematopoietic neoplasms and non-Hodgkin's lymphoma; the exposure-response relationship between cumulative exposure and leukemia was positive, although not statistically significant. Of interest is the additional observation that none of the other measures of exposure (i.e., duration, average and maximum) in this analysis were predictors of hematopoietic cancers, consistent with results of other investigations. Difficult to explain, though, in the context of causality is the decrease in hematopoietic cancer in women versus the increase in men observed in this cohort.

Therefore, the available epidemiological studies of the association between hematological cancers and exposure to ethylene oxide in occupationally exposed human populations fulfil, in part only, some of the traditional criteria for causality, including exposure-response and temporal relationship. Strength of the association is weak, although this may be a function of inadequate length of follow-up in existing studies. The observation of variations in response among males and females in the largest cohort study with the most extensive exposure analyses begs, to some extent, coherence.

Clearly, therefore, epidemiological evidence for the association between exposure to ethylene oxide and hematological cancers is not convincing in its own right. Assessment of the weight of evidence for carcinogenicity in human populations should not, however, be considered in isolation from the extensive supporting data on carcinogenicity, genotoxicity and inter- and intraspecies variations in metabolism and response.

Cytogenetic changes (i.e., increased frequency of chromosomal aberrations, micronuclei or sister chromatid exchange) within the peripheral blood cells have been reported in a number of cross-sectional studies, principally of populations exposed occupationally to ethylene oxide (Table 4). None of these studies is convincing in its own right, and inherent limitations of cross-sectional investigations make them less reliable than cohort or case-control studies as a basis for inference of causality. Nevertheless, observation of cytogenetic effects in some groups of workers exposed to elevated concentrations of ethylene oxide in the most sensitive studies, while not necessarily an indicator of chronic adverse health outcomes, provides some limited additional supporting evidence for the ability of ethylene oxide to interact with the genome in individuals exposed to this substance. Indeed, in comparison with that for other substances, the relative consistency of the data across studies is rather striking, although there are some inconsistent observations within studies, particularly in relation to the nature of clastogenic effects observed at various time points and exposures. For example, while not well characterized in individual studies, the increased occurrence of cytogenetic changes has tended to be observed at exposures to ≥5 ppm (9.2 mg/m³) ethylene oxide, thereby satisfying the criterion of exposure-response. In addition, while the numbers of subjects were relatively small in some of the investigations, results were consistently positive in the more sensitive (i.e., larger) studies of populations exposed to higher concentrations (i.e., those with >25 subjects, such as at site III in Galloway et al. [1986] and Stolley et al. [1984], high-dose group in Mayer et al. [1991], Ribeiro et al. [1994] and Richmond et al. [1985]). While there was inadequate control for confounding in several of the particularly small, early investigations (Garry et al. 1979; Yager et al. 1983), frequencies of clastogenic effects were sufficiently elevated in some cases that they were unlikely to be due to potential confounders (Laurent, 1988). It should also be noted that there was no information on genotype with respect to GSTT1 for the populations examined in these studies.

There is also overwhelming evidence for the biological plausibility of the carcinogenicity of ethylene oxide in human populations based upon data from carcinogenesis bioassays and other laboratory studies. An increased incidence of leukemias (although mononuclear) in F344 rats and lymphomas in mice (in addition to other types of tumours in both species) has been observed following inhalation of ethylene oxide (Lynch et al. 1984a,b; Snellings et al. 1984b; Garman et al. 1985; Garman and Snellings, 1986; NTP, 1987). Available data are insufficient to support a plausible mode of induction of tumours, however. While the spectrum of tumours induced in rats and mice is consistent (in part) with variations between species and tissues in the detoxification of the compound by the GSTT1 pathway, there is no correla tion w ith id entifi ed putatively responsible DNA adducts. However, the genotoxicity of ethylene oxide undoubtedly plays a critical role in tumour induction. Ethylene oxide is a potent alkylating agent that has been genotoxic in almost all available studies in laboratory animals. Gene mutations, DNA damage and cytogenetic effects have been observed routinely in bacterial, rodent and human cells exposed in vitro to ethylene oxide and in the somatic cells of laboratory species exposed in vivo to this substance.

Therefore, there is suggestive but inconclusive evidence (possibly attributable to the limitations of the studies) for an association between exposure to ethylene oxide and hematological cancers in occupationally exposed populations. There is rather consistent evidence that ethylene oxide interacts with the genome of cells within the circulatory system in occupationally exposed humans and overwhelming supporting evidence of biological plausibility based on carcinogenicity and genotoxicity in laboratory animals. Based on these considerations and the lack of qualitative differences in metabolism between humans and laboratory animals, ethylene oxide is considered highly likely to be carcinogenic to humans.

3.3.2.2 Heritable mutations

Although relevant data in humans are not available, dominant lethal mutations, heritable translocations, chromosomal aberrations, DNA damage and adduct formation in rodent sperm cells have been observed in a number of studies involving the exposure of rats and mice to ethylene oxide. Based upon the likely role for DNA alkylation in production of the genotoxic effects in germ cells in laboratory animals exposed to ethylene oxide, as well as the lack of qualitative differences in the metabolism of this substance between humans and animals (including DNA adduct formation), ethylene oxide can be considered a likely human germ cell genotoxicant.

3.3.2.3 Other non-neoplastic effects

3.3.2.3.1 Effects in humans

In humans, ethylene oxide vapour is irritating to the eyes, nose and throat. Aqueous solutions of ethylene oxide can be irritating to the skin, and, in some individuals, dermal irritation was associated with contact with ethylene oxide-sterilized materials and clothing. Ethylene oxide is considered an effective sensitizing agent (Bommer and Ritz, 1987; Bousquet and Michel, 1991). Type I (anaphylaxis) and Type IV (contact dermatitis) hypersensitivity reactions have been observed in individuals exposed to ethylene oxide. Anaphylactic reactions (ranging from mild to severe) have been noted among patients undergoing various forms of dialysis involving equipment sterilized by exposure to ethylene oxide. Asthmatic reactions may occur either alone or in combination with anaphylactic events, and case reports of occupational asthma attributed to ethylene oxide exposure have appeared (Dugue et al. 1991; Verraes and Michel, 1995).

Neurological effects (including neurophysiological, neurobehavioural and histopathological effects) have been clearly documented in workers exposed to relatively high concentrations of ethylene oxide. These include effects related to sensorimotor polyneuropathy, cognition, language and speech disturbances and seizures. Other effects attributed to ethylene oxide have included numbness and weakness in the extremities, slow and clumsy alternating hand movements, decreased muscle stretch reflexes in extremities, heel-shin ataxia, unsteady and wide-based gait, reduced ankle reflexes, diminished or impaired psychomotor skills and reduced peripheral nerve conduction velocity. Amelioration of symptoms following the cessation of exposure has been commonly observed. In individuals exposed to >700 ppm (1281 mg/m³) ethylene oxide, sural nerve biopsies revealed axonal degeneration with mild changes in the myelin sheath; muscle biopsies revealed degeneration atrophy (Kuzuhara et al. 1983).

Evidence from epidemiological studies of reproductive effects of ethylene oxide in humans is considered to be limited, at best, with suggestive but inconclusive evidence, at present, of increased risks of spontaneous abortion in female hospital workers (Hemminki et al. 1982, 1983) and dental assistants (Rowland et al. 1996) exposed to ethylene oxide in sterilization of equipment. There is also a single report of increased risk of spontaneous abortion in one investigation of women with partners who had had some potential for exposure to this substance (Lindbolm et al. 1991). While there are some consistent results in this regard, the available data are too limited to address other traditional criteria for causality such as strength and dose-response. With respect to biological plausibility, the data are supported, at least to some extent, by studies in animals that indicate that, among non-neoplastic effects, reproductive effects occur at lowest concentration.

Hematological changes (e.g., hematocrit, hemoglobin, lymphocytes, neutrophils) among workers occupationally exposed to higher concentrations of ethylene oxide have also been reported in some studies (Deschamps et al. 1990; Schulte et al. 1995). An increased prevalence of cataracts was noted among a small number of French hospital workers exposed to ethylene oxide (Deschamps et al. 1990).

3.3.2.3.2 Effects in laboratory animals

In laboratory animals, ethylene oxide is moderately acutely toxic. Data on the non-neoplastic effects following repeated exposure are somewhat limited due to emphasis in past studies on the carcinogenicity of the compound. However, in available studies, ethylene oxide has induced a wide range of effects in laboratory animals, including those at the site of contact and those on the hematological, reproductive and neurological systems. Effects on the neurological and reproductive systems occur at lowest concentrations.

With respect to neurological effects, histological alterations in the axons within the nucleus gracilis of the medulla oblongata and demyelination of the distal portion of the fasciculus gracilis within the medulla of monkeys following long-term exposure (Sprinz et al. 1982; Lynch et al. 1984b) and abnormal gait and reduced locomotor activity in mice after subchronic exposure (Snellings et al. 1984a) have been observed at lowest concentration. At higher concentrations, there is a wide range of more severe effects in rats, including awkward or ataxic gait, reversible paralysis and atrophy of the muscles of the hind limbs, accompanied, in some cases, by pathological evidence of axonal degeneration of myelinated fibres in nerves of the hind legs.

Reproductive effects in males observed in repeated-dose studies have included alterations in sperm morphology in rats (Mori et al. 1991), changes in sperm count and motility in monkeys at lower concentrations and degeneration of the seminiferous tubules and reductions in epididymal and testicular weights in rats at higher concentrations. In reproductive studies, reductions in litter size and increased post-implantation losses in rats are observed at lowest concentrations.

Ethylene oxide is fetotoxic in the presence and absence of maternal toxicity at concentrations higher than those associated with adverse reproductive and neurological effects, but it is teratogenic only at very high concentrations.

3.3.3 Exposure-response analysis

Inhalation is the principal route of exposure of the general population to ethylene oxide in the vicinity of industrial sources (i.e., those controllable under CEPA 1999). Moreover, virtually all of the available toxicological and epidemiological data relate to effects following exposure via this route. Indeed, information on exposure-response for ingestion of ethylene oxide is limited to reports from two (including one very early) studies in rats, in which gastric irritation and liver damage or histopathological changes in the stomach were observed in animals administered 100 mg ethylene oxide/kg-bw, 5 times per week for a total of 15 doses in 21 days (Hollingsworth et al. 1956), or 7.5 mg ethylene oxide/kg-bw twice weekly for 150 weeks (Dunkelberg, 1982), respectively. The remainder of this section, therefore, addresses exposure-response for the inhalation of ethylene oxide.

While the metabolism of ethylene oxide appears to be qualitatively similar in humans and animals, quantitative variations have not been well characterized. Two physiologically based pharmacokinetic (PBPK) models for ethylene oxide have been developed and verified for the rat (Hattis, 1987; Krishnan et al. 1992), although they have not been scaled to humans. Although subsequent studies have provided data that will ultimately be used in more refined PBPK models (Brown et al. 1996), publishe d report s of an updated model (including those for other species) have not been identified.

The parent compound is the putatively toxic entity, and exposures of the same concentration and duration are expected to result in equivalent toxicity across species. This is supported by the observed similarity in response to identical levels of exposure in the carcinogenicity bioassays in rats and mice. Therefore, no interspecies scaling to account for variations between inhalation rate to body weight ratios or body surface areas of humans to animals have been incorporated in the measures of dose-response reported here.

3.3.3.1 Carcinogenicity

Cancer is considered the critical endpoint for quantitation of exposure-response for risk characterization for ethylene oxide.⁴ This is based on the observation that tumours (and somatic mutations) are the effects that occur at lowest concentration. A statistically significant increased incidence of brain tumours was observed at concentrations as low as 60.4 mg/m³ in rats; moreover, incidences of several types of tumours were increased, although not significantly, at 18,3 mg/m³. Moreover, the genotoxicity of ethylene oxide, for which the weight of evidence is consistent and convincing, undoubtedly plays a critical role in tumour induction.

Quantitation of exposure-response for cancer for ethylene oxide is based on studies in laboratory animals, since limitations of the existing epidemiological data prevent adequate consideration of traditional criteria for causality (particularly with respect to periods of follow-up in the largest investigations). Moreover, available data indicate that the metabolism and mode of action of ethylene oxide in humans and laboratory animals do not differ qualitatively.

Data suitable for analysis of exposure-response are available from two carcinogenesis bioassays in F344 rats (Lynch et al. 1984a,b; Snellings et al. 1984b; Garman et al. 1985; Garman and Snellings, 1986) and one in B6C3F₁ mice (NTP, 1987). In F344 rats, there were dose-related increases in the incidence of mononuclear leukemias, peritoneal mesotheliomas and brain tumours; in mice, the incidence of lung carcinomas, malignant lymphomas, uterine adenocarcinomas, mammary carcinomas, adenosquamous carcinomas and Harderian cystadenomas was increased.

Concentrations of ethylene oxide causing a 5% increase in tumour incidence over background (i.e., Tumorigenic Concentration₀₅s, or TC₀₅s) were calculated by first fitting the multistage model to the dose-response data (see Figure 2). The multistage model is given by:

where d is dose, k is the number of dose groups in the study minus one, P(d) is the probability of the animal developing a tumour at dose d and qi > 0, i = 1,..., k are parameters to be estimated.

The models were fit using GLOBAL82 (Howe and Crump, 1982), and the TC₀₅s were calculated as the concentration C that satisfies:

A chi-square lack of fitness test was performed for each of the three model fits. The degrees of freedom for this test are equal to k minus the number of q_i's for which estimates are non-zero. A p-value less than 0.05 indicates a significant lack of fit.

The TC₀₅s and the corresponding 95% lower confidence limit (95% LCL) were adjusted for continuous exposure by multiplying the values by either 7/24 x 5/7 (for the study reported by Lynch et al. [1984a,b], in which animals were exposed for 7 hours per day, 5 days per week) or 6/24 x 5/7 (for the studies reported by Snellings et al. [1984b], Garman et al. [1985], Garman and Snellings [1986] and NTP [1987], in which animals were exposed for 6 hours per day, 5 days per week). Model parameters, the adjusted TC₀₅s and corresponding 95% LCLs are presented in Table 6,

For the tumours in rats, characterization of exposure-response was optimal in the study reported by Snellings et al. (1984b), Garman et al. (1985) and Garman and Snellings (1986). The number of dose groups was greatest in this bioassay, and two of the three doses were in a lower concentration range than in the study by Lynch et al. (1984a,b) (0, 18,3, 60.4 or 183 mg/m³ versus 0, 92 or 183 mg/m³). Dose spacing was excellent (approximately 3-fold variation between concentrations), both sexes were exposed and group sizes were slightly larger (120 per sex per group) than in the bioassay of Lynch et al. (1984a,b) (80 males per group).

For the study in rats in which exposure-response was best characterized (Snellings et al. 1984b; Garman et al. 1985; Garman and Snellings, 1986), the TC₀₅s range from 2.2 mg/m³ (95% LCL = 1.5 mg/m³) for mononuclear leukemia ⁵ in F344 rats to 31,0 mg/m³ (95% LCL = 16,1 mg/m³) for brain tumours. TC₀₅s for comparable tumours in the study in which exposure-response was less well characterized (Lynch et al. 1984a,b) were somewhat higher (12,5-31,9 mg/m³, respectively).

Values of the TC₀₅s in mice ranged from 6,7 mg/m³ (95% LCL = 4.2 mg/m³) for Harderian cystadenomas in males to 22.7 mg/m³ (95% LCL = 11.4 mg/m³) for uterine adenocarcinomas.

Figure 2 TC₀₅s for ethylene oxide

Enlarge image

Table 6 TC₀₅s for ethylene oxide

Tumour incidence	TC05 (mg/m3)	LCL on TC05 (mg/m3)	Chi- square	df	p-value
Male rats exposed to 0.92 or 183 mg ethylene oxide/m3, 7 hours/day, 5 days/week (Lynch et al., 1984a,b)1
Incidence of mononuclear cell leukemia: 24/77, 38/70, 30/76 Incidence of peritoneal mesothelioma: 3/78, 9/79, 21/79 Incidence of brain mixed cell glioma: 0/76, 2/77, 5/79	12.5 14.4 31.9	5.1 6.1 18.3	3.5 0 0	1 0 1	0.06 - 1.0
Male and female rats exposed to 0, 18,3, 60.4 or 183 mg ethylene oxide/m3, 6 hours/day, 5 days/week (Snellings et al., 1984b; Garman et al., 1985;Garman and Snellings, 1986) 2
Incidence of mononuclear leukemia in males: 13/97, 9/51, 12/39, 9/30 Incidence of mononuclear leukemia in females: 11/116, 11/54, 14/48, 15/26 Incidence of peritoneal mesothelioma in males: 2/97, 2/51, 4/39, 4/30 Incidence of primary brain tumours in males: 1/181, 1/92, 5/85, 7/87 Incidence of primary brain tumours in females: 1/188, 1/94, 3/92, 4/80	6,0 2.2 10.8 17.5 31.0	3.1 1.5 5.6 10.8 16.1	2.2 0.58 0.78 1.6 0.45	2 2 2 2 2	0.34 0.75 0.68 0.50 0.80
Male and female mice exposed to 0, 92 or 183 mg ethylene oxide/m3, 6 hours/day, 5 days/week (NTP, 1987) 2
Incidence of lung carcinoma in males: 6/50, 10/50, 16/50 Incidence of lung carcinoma in females: 0/49, 1/48, 7/49 Incidence of ma lignant ly mphoma in females: 9/49, 6/48, 22/49 Incidence of uterine aden ocarcinoma: 0/49, 1/47, 5/47 Incidence of mammary adenocarcinoma and adenosquamous carcinoma in females: 1/49, 8/48, 6/49 Incidence of Harderian cystadenoma in males: 1/43, 9/44, 8/42 Incidence of Harderian cystadenoma in females: 1/46, 6/46, 8/47	10.2 19.8 12,2 22.7 10.4 6.7 9.1	4.1 10.3 6.3 11.4 6.0 4.2 5.5	0 0.34 3.5 0.07 3.0 2.0 0.30	0 2 1 2 1 1 1	- 0.84 0.06 0.97 0.08 0.16 0.58

It should be noted, however, that characterization of exposure-response in the NTP (1987) bioassay on which these values are based was not optimal; there were only two dose groups and controls, with the lowest administered concentration being 92 mg/m³.

For none of the modelled TC₀₅s was there significant lack of fit (p > 0.05, Table 6). For the study in rats in which exposure-response was best characterized (Snellings et al. 1984b; Garman et al. 1985; Garman and Snellings, 1986) and that in mice (NTP, 1987), fits for malignant lymphomas and mammary adenocarcinomas and adenosquamous carcinomas (combined) in females in the latter investigation were poorest (p = 0.06 and 0.08, respectively).

Based on modelling (using THC program; Howe, 1995) of the incidence of Hprt mutations in splenic T-lymphocytes of male B6C3F₁ mice (Big BlueÒ, lacI transgenic) exposed to ethylene oxide for 4 weeks ⁶ (Walker et al. 1997a), the Benchmark Concentration₀₅ (BMC₀₅) for somatic cell mutations (i.e., the concentration associated with a 5% increase in the incidence of Hprt mutation) (adjusted for intermittent to continuous exposure) was within the range of the lowest TC₀₅s in rats and mice. It should be noted, however, that characterization of exposure-response in Walker et al. (1997a) was not optimal; although there were three dose groups and controls, the lowest administered concentration was 92 mg/m³.

In the interest of utilizing all available data to inform characterization of exposure-response, the tumorigenic potencies developed based on studies in animals were compared with risks of hematological cancers reported in epidemiological studies in populations occupationally exposed to ethylene oxide. The protocol and results of these analyses are reported elsewhere (Health Canada, 1999b). Results indicated that risks predicted based on the most sensitive outcome in rats (mononuclear cell leukemia in female F344 rats) were consistent with the confidence intervals of the SMRs observed for both leukemias overall and all hematopoietic neoplasms in males in the cohort study by Stayner et al. (1993) (i.e., the only epidemiological study in which individual cumulative exposure was characterized). However, the limitations of this comparative exercise preclude its meaningful contribution to quantitation of risk. These include uncertainties of the available epidemiological data on ethylene oxide, which prevent adequate consideration of traditional criteria for causality (particularly with respect to periods of follow-up in investigations of greatest sensitivity). Moreover, meaningful direct comparison of potency in laboratory animals with that in humans is precarious at best, in light of the inadequacy of available information on interspecies variations in kinetics and metabolism and mode of action to serve as a basis for characterization of site concordance between animals and humans and the extremely wide range of the confidence limits on the SMRs in the epidemiological studies.

3.3.3.2 Heritable mutations

There have been several efforts to quantify the genetic risk to the offspring of humans exposed to ethylene oxide, the most comprehensive of which is that of Natarajan et al. (1995), which documents the outcome of deliberations of an international workshop of experts. This exercise was undertaken to identify data gaps that would permit a more refined estimate of heritable genetic risk from ethylene oxide and to acquire experience with the parallelogram approach to better inform future efforts in this area. The outcome is presented here primarily as a basis for comparison with the tumorigenic potencies for cancer, to ensure that measures developed for this endpoint will be protective for other reported effects. However, it meets this objective only in part, since the calculated genetic risk is underestimated, is based on induced dominant visible mutations only and does not take into consideration recessive mutation, dominant lethal mutations or heritable translocations. The relevant data for these endpoints were judged either not to be sufficiently robust or to result in a very small increment in actual genetic risk to live offspring. An increase in dominant lethal mutations in humans might be manifested in an increase in spontaneous abortions, as reported in some hospital sterilization workers (Hemminki et al. 1982).

The analysis was based on induced dominant visible mutations in mice from a study by Lewis et al. (1986), which was designed to mimic human occupational exposure (i.e., involving exposure for prolonged periods in order to cover all stages of spermatogenesis). Using the parallelogram approach and additional quantitative data on somatic mutations (Hprt in splenocytes) in mice (Walker et al. 1994) and in an occupationally exposed human population (HPRT) (Tates et al. 1991), it was estimated that exposure for one working year (1800 hours) to 1 ppm ethylene oxide would lead to an incremental risk of 4 x 10^-4 above background that a disease with dominant inheritance would be transferred to the offspring. As a basis for comparison with the potency estimates for cancer, the BMC₀₅ for this effect would be 46 mg/m³.⁷

Identified sources of uncertainty of the estimate were the doubling dose for Hprt mutations in the mouse, the doubling dose for HPRT mutations in humans, the mutation rate in mice, the number of loci involved, the risk from exposure of females, the extrapolation from mutation frequency to dominant disease states and possible influence of dose rates (Natarajan et al. 1995). Although there was some attempt by the authors to quantitate uncertainty from these sources, such an estimate does not reflect uncertainty associated with reliance on limited (possibly unrepresentative) data, which could be considerably greater.

3.3.3.3 Other non-neoplastic effects

3.3.3.3.1 Humans

Exposure-response for the neurological effects (including neurophysiological, neurobehavioural and histopathological effects) observed in workers exposed to ethylene oxide has not been well characterized. In case studies, reported levels of ethylene oxide have ranged from 4.2 to >700 ppm (7.7 to >1281 mg/m³) (Gross et al. 1979; Salinas et al. 1981; Finelli et al. 1983; Kuzuhara et al. 1983; Zampollo et al. 1984; Schroder et al. 1985; Fukushima et al. 1986; Ristow and Cornelius, 1986; Crystal et al. 1988). In surveys, typical TWA exposures have ranged from <1 to 4.7 ppm (<1.8 to 8.6 mg/m³), with peaks as high as 250 ppm (458 mg/m³) ethylene oxide (Estrin et al. 1987, 1990; Klees et al. 1990). In individuals exposed to >700 ppm (1281 mg/m³) ethylene oxide, sural nerve biopsies revealed axonal degeneration with mild changes in the myelin sheath; muscle biopsies revealed degeneration atrophy (Kuzuhara et al. 1983).

Available data on other potential effects in humans associated with exposure to ethylene oxide (e.g., hematological, ocular and reproductive/developmental) are limited and inadequate for characterization of exposure-response.

3.3.3.3.2 L aboratory an imals

Although ethylene oxide has induced a wid e range of no n-neoplastic effects in laboratory animals, it is those on the neurological and reproductive systems that occur at lowest concentrations.

With respect to neurological effects, abnormal gait and reduced locomotor activity (in two of five tests) were observed in small numbers of mice (n = 5) after subchronic exposure to ethylene oxide at concentrations ranging from 86 to 425 mg/m³ (Snellings et al. 1984a); at the highest concentration (425 mg/m³), there were significant differences for toe and tail pinch reflexes and abnormal righting reflex. In another study in mice, however, there were no similar clinical signs (although neurobehavioural changes were not addressed specifically) in animals exposed to 50 or 100 ppm (92 or 183 mg/m³) ethylene oxide for 2 years (NTP, 1987). In monkeys, histological alterations in the axons within the nucleus gracilis of the medulla oblongata and demyelination of the distal portion of the fasciculus gracilis within the medulla were observed following long-term exposure to ≥92 mg ethylene oxide/m³ (Sprinz et al. 1982; Lynch et al. 1984b).

Effects of ethylene oxide on the testes of males in repeated-dose toxicity studies are the reproductive effects observed at lowest concentration, although there is a preliminary (unverifiable) report that there have been reductions in litter size and increased post-implantation losses in the F₀ animals at somewhat lower levels. Alterations in sperm morphology in rats (Mori et al. 1991) and changes in sperm count and motility in monkeys (Lynch et al. 1984c) were observed following exposure to 92 mg ethylene oxide/m³. In rats, reductions in some reproductive parameters (i.e., number of pups born per litter, number of implantation sites per female) were observed in animals exposed before mating and during gestation to 100 ppm (183 mg/m³) ethylene oxide (Snellings et al. 1982b). Reportedly, reductions in litter size and increased post-implantation losses were observed in the F₀ and F₁ generations of rats exposed to 100 ppm (183 mg/m³) ethylene oxide, although a full account is not yet available; these effects were also noted in F₀ animals exposed to 33 ppm (60.4 mg/m³) ethylene oxide (Snellings, 1999).

Developmental effects that occurred at lowest concentrations were reductions in fetal body weight, without effects on skeletal length or ossification, following exposure of the dams to 100 ppm (183 mg/m³) ethylene oxide during gestation (Snellings et al. 1982a); this concentration appeared to have no overt effects on the dams.

In rats exposed chronically to ethylene oxide (Snellings et al. 1984b), a slight (unspecified) reduction in body weight gain (in females) was reported at 33 ppm (60.4 mg/m³). (There were only a limited number of non-neoplastic endpoints observed in this study.)

A slight increase in the prevalence of lens opacities has also been observed in monkeys exposed for 2 years to ≥92 mg ethylene oxide/m³ (Lynch et al. 1992). Hematological effects have also been observed in rats, mice and dogs exposed for various periods to greater concentrations of ethylene oxide than those addressed here (Jacobson et al. 1956; Popp et al. 1986; Katoh et al. 1988, 1989; Fujishiro et al. 1990; Mori et al. 1990).

Based on available documented studies, therefore, non-neoplastic effects of ethylene oxide have been observed only at concentrations greater than those at which increases in tumours have been reported in other studies (i.e., the latter was observed at concentrations as low as 18,3 and 60.4 mg/m³ in rats). In addition, in view of the likely critical role of the genotoxicity of ethylene oxide, for which the weight of evidence is consistent and convincing in the induction of tumours, cancer is clearly the critical endpoint for quantitation of exposure-response for risk characterization, and measures based on this endpoint will be protective for other reported effects. For example, a Tolerable Concentration based upon observed effects on the sperm (i.e., reduced count and motility) and brain (i.e., nerve dystrophy and demyelination) in monkeys exposed chronically to ethylene oxide (Sprinz et al. 1982; Lynch et al. 1984b,c; Setzer et al. 1996) or upon reproductive effects (i.e., reduced number of pups per litter and implantation sites per female) observed in rats exposed subchronically to ethylene oxide, prior to and during mating as well as during gestation (Snellings et al. 1982b), would be in the range of tens of µg/m³.

3.3.4 Human health risk characterization

For substances such as ethylene oxide, where there is a strong likelihood that the mode of action for the induction of tumours involves direct interaction with genetic material, estimates of exposure are compared with quantitative estimates of carcinogenic potency (Exposure Potency Index, or EPI) to characterize risk and to provide guidance in establishing priorities for further action (i.e., analysis of options to reduce exposure) under CEPA.

The lowest TC₀₅ in the study in rats with optimal characterization of exposure-response (Snellings et al. 1984b; Garman et al. 1985; Garman and Snellings, 1986) and in mice (NTP, 1987) was 2.2 mg/m³ for the development of mononuclear leukemias in F344 female rats exposed via inhalation to ethylene oxide; the 95% LCL was 1.5 mg/m³ (Table 6). The margins between carcinogenic potency and the extremely limited data on measured and predicted concentrations of ethylene oxide in ambient (and indoor) air in Canada (and elsewhere) are presented in the table below. On this basis, the priority for investigation of options to reduce exposure in the vicinity of point sources is considered to be high. However, it should be noted that this is based on concentrations modelled, taking into account information on releases, which have not been validated by monitoring data.⁸ Based upon margins between censored mean concentrations for monitoring data from a multimedia exposure study conducted in Canada, the priority for investigation of options to reduce exposure to ethylene oxide is moderate to high, although it should be noted that this is based on detection in a very small proportion of samples in the study.

Concentration of ethylene oxide	Potency (TC05) : [95% LCL] (2200 µg/m3) : [1500 µg/m3]	Margin between potency and concentration	Exposure Potency Index (EPI)	Priority for further action (Health Canada, 1994)
0.0062 µg/m3; concentration in ambient air in southern Ontario predicted from ChemCAN fugacity model	(2200) :[1500]	(350 000) : [240 000]	(2.9 x 10-6) : [4.2 x 10-6]	(Moderate) : [Moderate]
0.34 µg/m3; censored mean concentration in ambient air from multimedia survey in Canada (Health Canada, 1999a)	(2200) : [1500]	(6500) : [4400]	(1.5 x 10-4) : [2.3 x 10-4]	(Moderate) : [High]
0.17 µg/m3; censored mean concentration in indoor air from multimedia survey in Canada (Health Canada, 1999a)	(2200) : [1500]	(13 000) : [8800]	(7.7 x 10-5) : [1.1 x 10-4]	(Moderate) : [Moderate]
2.12 µg/m3; predicted maximum average daily concentration in ambient air in the vicinity of Canadian hospitals	(2200) : [1500]	(1040) : [710]	(9.6 x 10-4) : [1.4 x 10-3]	(High) : [High]
4.9 µg/m3; maximum concentration in ambient air from multimedia survey in Canada (Health Canada, 1999a)	(2200) : [1500]	(450) : [310]	(2.2 x 10-3) : [3.2 x 10-3]	(High) : [High]
4.0 µg/m3; maximum concentration in indoor air from multimedia survey in Canada (Health Canada, 1999a)	(2200) : [1500]	(550) : [375]	(1.8 x 10-3) : [2.7 x 10-3]	(High) : [High]
20.1 µg/m3; predicted maximum 1-hour ground-level concentration in ambient air in the vicinity of an ethylene oxide production facility in Canada	(2200) : [1500]	(110) : [75]	(9.1 x 10-3) : [1.3 x 10-2]	(High) : [High]
2 µg/m3; predicted maximum average annual concentration in ambient air in the vicinity of a hospital in Florida (Tutt and Tilley, 1993)	(2200) : [1500]	(1100) : [750]	(9.1 x 10-4) : [1.3 x 10-3]	(High) : [High]
11 µg/m3; predicted maximum average annual concentration in ambient air in the vicinity of a sterilization facility in Florida (Tutt and Tilley, 1993)	(2200) : [1500]	(200) : [140]	(5,0 x 10-3) : [7.1 x 10-3]	(High) : [High]

3.3.5 Uncertainties and degree of confidence in human health risk characterization

There are considerable uncertainties in the assessment of human exposure to ethylene oxide resulting from the paucity of information concerning current levels in media to which the general population of Canada is currently exposed. There is a high degree of certainty that ethylene oxide has been and is being discharged to the atmosphere in Canada from chemical manufacturing facilities and from hospital sterilizers. There is a moderate degree of certainty that these emissions have been decreasing in recent years, but this conclusion is based on voluntary reporting of emission estimates and has not been validated by comparison of historic and current data concerning concentrations in the outdoor air in the vicinity of point sources of atmospheric discharge. In addition, the potential impact of a new ethylene oxide/glycol production facility in Scotford, Alberta, on trends of future releases is not known. There is a high degree of uncertainty regarding the range of concentrations of ethylene oxide in the atmosphere or in air in the vicinity of point sources in Canada, as relevant monitoring data have not been identified. Rather, estimates included herein are restricted to those based on unvalidated fugacity and dispersion modelling.

There is a very high degree of uncertainty regarding concentrations of ethylene oxide in the indoor air of Canadian residences and public places. Other than environmental tobacco smoke, potential indoor sources of ethylene oxide have not been identified. There is a moderate degree of certainty that smokers have higher daily intakes of ethylene oxide than non-smokers, but data on the amounts of ethylene oxide in the mainstream and sidestream smoke of Canadian cigarettes were not identified. There is a moderate degree of certainty that ethylene oxide is not released in significant amounts from consumer products in which this substance is incorporated during manufacture.

There is a moderate degree of certainty that the consumption of drinking water does not contribute significantly to the intake of ethylene oxide in Canada. Although no data were identified concerning concentrations in surface water, groundwater or drinking water in Canada, ethylene oxide has been detected only very infrequently in water in the United States. The physical/chemical properties of ethylene oxide and the fact that it is released to the atmosphere support the conclusion that concentrations in water in Canada would be negligible.

There is a very high degree of uncertainty concerning the levels of ethylene oxide in foods consumed in Canada, since relevant monitoring data were not identified. There is a high degree of certainty that among potential food sources, spices are most likely to have the highest concentrations, as fumigation of spices with ethylene oxide is permitted in Canada. Monitoring of ethylene oxide concentrations in Canadian foodstuffs is clearly desirable to improve estimates based on limited and possibly unrepresentative data on levels of ethylene oxide in foodstuffs from other countries, and highly uncertain values for consumption rates indicate that food may be a significant source of exposure.

The degree of confidence in the database on the toxicity of ethylene oxide is moderate. While the database on non-cancer toxicity in laboratory animals is limited, there is a high degree of confidence that cancer and heritable genotoxicity occur at lowest concentrations, and risk management measures developed on the basis of exposure-response for these effects will be protective for other adverse effects in the general population.

The carcinogenicity of ethylene oxide in humans has been investigated in a number of studies, the largest of which involved a cohort of more than 18 000 individuals. However, limitations of these investigations prevent adequate consideration of traditional criteria for causality (particularly with respect to periods of follow-up in investigations of greatest sensitivity). Similarly, epidemiological studies on cytogenetic changes and reproductive effects in human populations are inadequate to allow any inference concerning causality to be drawn.

While there is a high degree of certainty that the genotoxicity of ethylene oxide plays an important role in the carcinogenicity of this substance and that the metabolism and mode of action of ethylene oxide in humans and laboratory animals do not differ qualitatively, the mode of action in inducing cancer or heritable genotoxic effects has not been clearly delineated. Possible quantitative variations between humans and animals have also not been elucidated.

Meaningful direct comparison of carcinogenic potency in laboratory animals with that in humans is precluded due to limitations of the epidemiological database, the inadequacy of available information on interspecies variations in kinetics, metabolism and mode of action to serve as a basis for characterization of site concordance between animals and humans, and the extremely wide range of the confidence limits on the SMRs in the epidemiological studies.

There is some uncertainty concerning the relevance to humans of mononuclear cell leukemias in F344 rats, since this type of tumour is specific to this strain of rat, it arises spontaneously with a significant frequency in older unexposed animals, and its etiology has not been definitively identified. However, TC₀₅s for the tumours with next greatest potency for the studies in rats and mice with optimum characterization of exposure-response and resulting EPIs would be approximately only 3-fold greater, and priorities for further action would remain the same. The 95% LCL on the TC₀₅ for mononuclear leukemia in female rats was 1.5 mg/m³, versus the maximum likelihood estimate of 2.2 mg/m³. Based upon the highest TC₀₅ identified from the study in which exposure-response was best characterized (i.e., 31,0 mg/m³ for primary brain tumours in female F344 rats), the resulting EPIs would be approximately 14-fold lower than those derived (in Section 3.3.4) on the basis of the mononuclear cell leukemias in female F344 rats.

3.4 Conclusions

CEPA 1999 64(a): Based on available data, it is unlikely that ethylene oxide is entering or may enter the environment in a quantity or concentration or under conditions that have or may have an immediate or long-term harmful effect on the environment or its biological diversity. Therefore, ethylene oxide is not considered to be "toxic" as defined under Paragraph 64(a) of CEPA 1999.

CEPA 1999 64(b): Based on available data, it is unlikely that ethylene oxide is entering or may enter the environment in a quantity or concentration or under conditions that constitute or may constitute a danger to the environment on which life depends. Therefore, ethylene oxide is not considered to be "toxic" as defined under Paragraph 64(b) of CEPA 1999.

CEPA 1999 64(c): Based on available data, it has been concluded that ethylene oxide is entering the environment in a quantity or concentration or under conditions that constitute or may constitute a danger in Canada to human life or health. Therefore, ethylene oxide is considered to be "toxic" as defined under Paragraph 64(c) of CEPA 1999. This approach is consistent with the objective that exposure to compounds where induction of cancer through direct interaction with genetic material cannot be ruled out be reduced wherever possible and obviates the need to establish an arbitrary "de minimis" level of risk for the determination of "toxic" under CEPA 1999. On the basis of limited monitoring data and predicted concentrations of ethylene oxide in air, the priority for investigation of options to reduce exposure, particularly in the vicinity of point sources, is considered to be high.

Overall conclusion: Based on critical assessment of relevant information, ethylene oxide is considered to be "toxic" as defined in Section 64 of CEPA 1999.

3.5 Considerations for follow-up (further action)

Based on comparison of extremely limited monitoring data and primarily predicted concentrations of ethylene oxide in air with tumorigenic potency, it is recommended that options to reduce exposure, particularly in the vicinity of point sources, be investigated. It is also recommended that there be additional investigation of the magnitude of exposure of populations in the vicinity of point sources to assist risk management actions.