Health Canada
www.hc-sc.gc.ca
Common menu bar links
Priority Substances List - Statement of the Science Report for Ethylene Oxide
2.0 Summary of Information Critical to Assessment of "Toxic" under CEPA 1999
2.1 Identity and physical/chemical properties
Ethylene oxide is also known as diethylene oxide, E.O., epoxyethane, 1,2-epoxyethane, oxane, oxidoethane and oxirane. Its Chemical Abstracts Service (CAS) registry number is 75-21-8. The chemical structure of ethylene oxide is shown in Figure 1.
pressure, ethylene oxide is a colourless, highly reactive and flammable gas with a characteristic ethereal odour. It has a high vapour pressure (~146 kPa) and high water solubility (completely miscible). It is very reactive in both the liquid and vapour phases (WHO, 1985). Table 1 summarizes the physical and chemical properties of ethylene oxide.
Figure 1 Chemical structure of ethylene oxide
The molecular formula for ethylene oxide is is H2COCH2 , and its molecular weight is 44.05. At room temperature (25°C) and normal atmospheric pressure, ethylene oxide is a colourless, highly reactive and flammable gas with a characteristic ethereal odour. It has a high vapour pressure (~146 kPa) and high water solubility (completely miscible). It is very reactive in both the liquid and vapour phases (WHO, 1985). Table 1 summarizes the physical and chemical properties of ethylene oxide.
The conversion factor for airborne ethylene oxide used throughout this report is 1ppm = 1.83 mg/m3.
2.2 Entry characterization
2.2.1 Production, uses and importation
Canadian companies that currently produce ethylene oxide include Dow Chemical Canada Ltd., Union Carbide Canada Inc. and Alberta & Orient Glycol. Domestic production was estimated at 625 kilotonnes in 1996, and the forecasted total estimated for 1999 was 682 kilotonnes (CPI, 1997). Dow Chemical Canada Ltd., in Alberta, had an estimated 60-kilotonne increase in production capacity between 1981 and 1992 and a forecasted 50-kilotonne increase between 1996 and 1999 (CPI, 1997). Alberta & Orient Glycol went into operation in September 1994 and has a potential capacity of 215 kilotonnes per year. Canadian import volumes for 1992, 1994 and 1996 were 4.39, 10.97 and 8.00 kilotonnes, respectively (CPI, 1997). Totals of 36 and 18 kilotonnes were exported in 1992 and 1996, respectively (CPI, 1997).
Shell Chemicals Canada intends to build a world-scale ethylene oxide/glycol plant at Scotford, Alberta, to help meet the growing worldwide demand for glycol. All of the ethylene oxide produced will be used in the synthesis of ethylene glycol, with an expected capacity of 400 kilotonnes per year. The new Scotford plant will supply customers in North America and the Far East starting in the year 2000 (CPI, 1997).
Virtually all of the ethylene oxide produced is used as an intermediate in the production of various chemicals (ATSDR, 1990). In 1993, 89% of the total Canadian production of ethylene oxide was used in the production of ethylene glycol (SRI, 1993); in 1996, 95% was used for this purpose (CPI, 1997). It is forecasted that the volume used in ethylene glycol production will continue to increase. An estimated 4% (26 000 tonnes) is used in the manufacture of surfactants (CPI, 1997). Ethylene oxide alone or in combination with other inert gases such as carbon dioxide and nitrogen is used to sterilize instruments from the health care, publication and wood products sectors. Ethylene oxide is also used in other industries where heat-sensitive goods are sterilized (How-Grant, 1991; BUA, 1995). In Canada, ethylene oxide is also used in the manufacture of choline chloride, glycol ethers and polyglycols (CPI, 1997). Other minor uses worldwide include use in the manufacture of rocket propellant and petroleum demulsifiers (Lewis, 1993).
Alternatives to ethylene oxide use in sterilization are available and include hydrogen peroxide, gas plasma, peracetic acid, ozone, chlorine dioxide, and low-temperature steam and formaldehyde. Some of the main advantages of using ethylene oxide as a sterilant are that it is very effective at penetrating most wrappings of paper and cloth, does not damage the materials or packaging and requires relatively low temperatures and pressures (BUA, 1995).
Ethylene oxide is used as an active ingredient in one registered pest control product in Canada for the control of insects in stored products and the control of bacteria in spices and natural seasonings. The quantity of ethylene oxide used as a fumigant in Canada is unknown (Ballantine, 1997). Ethylene oxide is also found as a formulant or component of a formulant in 25 other pest control products. In all 25 cases, amounts of ethylene oxide range from trace to 0.423%. The formulants include fungicides, insecticides, herbicides and an adjuvant (Ballantine, 1997). The use of ethylene oxide as an active ingredient in agricultural pesticides is not considered further in this assessment, as such use is regulated under the Pest Control Products Act.
2.2.2 Sources and releases
2.2.2.1 Natural sources
Ethylene oxide is known to be produced from a few natural sources. In certain plants, ethylene (a natural plant growth regulator) is degraded to ethylene oxide (Abeles and Dunn, 1985). It is also a product of ethylene catabolism in certain microorganisms (De Bont and Albers, 1976). Ethylene oxide can be generated from waterlogged soil (Smith and Jackson, 1974; Jackson et al., 1978), manure and sewage sludge (Wong et al., 1983). Quantitative estimates of production from these natural sources are not available, but emissions are expected to be negligible.
2.2.2.2 Anthropogenic sources
2.2.2.2.1 Non-point sources
There are several non-point sources of release of ethylene oxide. Safety precautions taken when handling and transporting the chemical, because of its explosive nature, seem to have reduced the incidence of spills. From the National Analysis of Trends in Emergency Systems database, only one spill was reported, and that was the release, due to human error, of 0.127 kg of ethylene oxide to soil at a government facility in Halifax, Nova Scotia (NATES, 1994). The CANUTEC database (Transport Canada, 1996) reported several incidents involving the release of ethylene oxide from derailed train tanker cars and truck accidents. Concentrations in the environment from these releases were not reported.
Other recognized non-point sources of release of ethylene oxide include its formation from fossil fuel combustion (U.S. EPA, 1984) and presence in tobacco smoke (Howard, 1989). Neither source is expected to be significant (U.S. EPA, 1984). Ethylene oxide is used as a component in the production of polyoxyethylene surfactants at high molar concentrations (5-20 mol/L) (Gaskin and Holloway, 1992). These surfactants increase the effectiveness of herbicide applications to plants; however, the ethylene oxide in this form is bound within the surfactant molecule, and any release is expected to be minimal. Similarly, ethylene oxide may be present in nonylphenol ethoxylate formulations at concentrations below 10 mg/L (Talmage, 1994), and, according to German manufacturers, ethylene oxide may remain as a contaminant at 0.001% or 10 mg/kg in liquid detergents from the production of non-ionic surfactants (BUA, 1995). Examination of the Material Safety Data Sheets for more than 15 different ethylene oxide-derived surfactants revealed final product concentrations at trace amounts (i.e., <1 ppm). A variety of other products, including paints and coatings, were reported to contain ethylene oxide at levels ranging from trace to <0.5%. The diffuse release of such quantities is expected to result in negligible environmental concentrations.
Ethylene oxide is used for the control of insect (i.e., fumigation) and microbial (i.e., sterilization) infestations. Estimates of the concentrations of ethylene oxide required for these purposes range from 250 to 1500 mg/L (Agriculture and Agri-Food Canada, 1996; Conviser, 1999; Health Canada, 1999a). Residues of ethylene oxide following the fumigation or sterilization of foods include unreacted ethylene oxide and various degradation products. Following fumigation, any residual ethylene oxide in foodstuffs disappears rapidly; concentrations generally fall to negligible levels within a few hours (IARC, 1976). However, degradation products of ethylene oxide (e.g., ethylene chlorohydrin) are generally less volatile and more persistent within foodstuffs. The presence and concentrations of such residues after fumigation or sterilization depend upon the concentration of ethylene oxide, temperature, aeration, storage conditions after treatment, the type of commodity, its moisture and lipid content and other factors (WHO, 1985). In Canada, ethylene oxide is permitted to be used (Food and Drugs Act, 1996) only as a food additive (fumigant) on "whole or ground spices except mixtures containing salt" at levels consistent with "good manufacturing practice" (i.e., the minimum amount required to achieve the intended effect). A legal tolerance of 1500 mg/kg has been established for spices (Health Canada, 1999a) with respect to the content of ethylene chlorohydrin, formed by the reaction of ethylene oxide with inorganic chlorides contained in the food. A proposal to establish a tolerance limiting residues of ethylene oxide used as a food additive in spices to 50 mg/kg is under consideration (Le Maguer, 1999).
2.2.2.2.2 Point sources
Gas and liquid forms of ethylene oxide can be released during production and use, as well as during the manufacture of ethylene glycol, ethoxylates, ethers and ethanolamines (Howard, 1989). Ethylene oxide releases to the Canadian environment as reported to the National Pollutant Release Inventory (NPRI) totalled 103.9 tonnes in 1993, 51.1 tonnes in 1994, 26.2 tonnes in 1995 and 22.8 tonnes in 1996 (NPRI, 1993, 1994, 1995, 1996). Between 1994 and 1995, there was an overall reduction of 49%, with the decline in releases attributable to Huntsman Corp. (decreased 88%), Dow Chemical Canada Inc. (decreased 66%) and Union Carbide Canada Inc. (decreased 41%). Twelve facilities were identified in 1994, 11 in 1995 and 10 in 1996, with all releases being to the atmospheric environment. Alberta and Ontario contributed 41% and 35% of the total releases, respectively, in 1995 (NPRI, 1995). The industrial sectors reporting releases of ethylene oxide include plastics and synthetics, inorganic chemicals (6.1 tonnes per year), industrial organic chemicals (8.7 tonnes per year) and soap and cleaning compounds (8.0 tonnes per year), with the quantities from the latter sector expected to diminish by 5.3 tonnes in 1997 (NPRI, 1996). An Environment Canada survey of Canadian industry regarding ethylene oxide use and release, which was carried out under the authority of Section 16 of the Canadian Environmental Protection Act, 1988 (CEPA 1988), revealed three additional facilities releasing a total of 3.5 tonnes in 1996. Two of these companies are involved in servicing medical facilities using ethylene oxide in sterilization processes and commercial sterilization operations and account for 86% (3.0 tonnes) of the release volume from this sector (Environment Canada, 1997b).
The Accelerated Reduction/Elimination of Toxics (ARET) program has reported emissions of 19 tonnes for 1997 by participating companies, with emissions reduced by 82% from the 1993 levels. Eight of the 10 facilities reporting to NPRI in 1996 are participating in ARET (ARET, 1999).
Ethylene oxide emissions from specific chemical production processes were estimated at 389 tonnes per year from an ethylene oxidation processing plant producing 210 000 tonnes per year under no control measures and reduced to 21.5 tonnes per year with the appropriate controls (Shen and Minns, 1997). Similarly, estimated emissions from an ethylene glycol manufacturing plant were 7.2 tonnes per year with no control devices and 0.36 tonnes per year with control devices (Shen and Minns, 1997). These estimates were slightly higher than releases reported from Dow Chemical Canada Inc. and Union Carbide Canada Inc., the two major ethylene oxide production facilities in Canada, which reported releases of 6.1 and 4.6 tonnes, respectively, in 1996 (NPRI, 1996).
An examination of the fugitive emissions from U.S. ethylene oxide production facilities in 1988 suggests that significant emissions can occur as a result of faulty design and inadequate maintenance or monitoring (Berglund et al., 1990). Data were collected from most of the ethylene oxide producers within the United States as part of a cooperative agreement between the Ethylene Oxide Industry Council, a trade association within the Chemical Manufacturers Association, and the U.S. Environmental Protection Agency (EPA). This study found that, overall, 86.9% of the components (i.e., gas valves, liquid valves, safety valves, pumps, flanges, open-ended lines, compressors) did not leak (<10 mg/m3), 10.3% were low emitters (100-1000 mg/m3), 0.6% were high emitters (>1000 mg/m3) and 0.6% were leakers (>10 000 mg/m3). Over 90% of the safety valves had no measurable level; however, leaking pumps and flanges accounted for 30 and 40% of the total estimated ethylene oxide emissions, respectively (Berglund et al., 1990).
Although sterilization is not a major use of ethylene oxide in terms of volumes consumed, it may be a very significant source of release to the environment (WHO, 1985). A survey of hospitals using ethylene oxide as a sterilizing agent was conducted in April 1994 by the Canadian Hospital Association (CHA and Environment Canada, 1994). Of the 204 hospitals contacted in all provinces and territories, 89 (44%) responded. A total of 70 (79%) of the respondents used ethylene oxide in a proportion of 12%, and the remaining 19 (21%) used it at 100%. In Canada, the amount of ethylene oxide used as a sterilant can be only roughly estimated. Based on the 1994 estimate of the volume of chlorofluorocarbon (CFC) used in sterilization facilities of 231 tonnes (Madé, 1996) and assuming that 79% of Canadian hospitals use the common ratio of ethylene oxide to carrier gas of 12:88 during sterilization operations (Meiners and Nicholson, 1988; CHA and Environment Canada, 1994), an estimated 31.5 tonnes of ethylene oxide were used in 1994 for sterilization at these facilities. Assuming that the remaining 21% of hospitals use it at 100% strength and that the amount of ethylene oxide per sterilization is equivalent to that used in the 12:88 method, an additional 8.4 tonnes are used, for a total of 39.9 tonnes per year. In the past, many hospitals did not have control equipment to reduce emissions; however, many facilities now have improved control measures in place (Havlicek et al., 1992; CHA and Environment Canada, 1994). In addition, owing to the adverse environmental impact of CFCs, many hospitals have begun to use alternative equipment that does not involve use of ethylene oxide (Smyth-Plewes, 1998). The current volumes of ethylene oxide used and released may be significantly less than the 1994 estimates.
The environmental control measures on sterilization equipment vary from virtually 100% control (CHA and Environment Canada, 1994) to no control (Markwordt, 1985). In the United States, more than 50% of the medical devices manufactured are sterilized with ethylene oxide, and every hospital performing surgery has at least one ethylene oxide sterilizer (How-Grant, 1991).
In an examination of the primary U.S. sources of ethylene oxide releases, sterilization/fumigation sites, production/captive use, medical facilities and ethoxylation account for 57, 31, 8 and 4% of total emissions, respectively (Markwordt, 1985). In an early U.S. study, it was estimated that <0.1% of ethylene oxide produced is used as a sterilizing agent or fumigant, yet this accounts for the majority of ethylene oxide released into the atmosphere (Markwordt, 1985). Similarly, Berkopec and Vidic (1996) found that, in Slovenia, emissions to the atmosphere during sterilization were higher than emissions from other processes, such as synthesis of glycols and other derivatives in the chemical industry, although the sterilization process accounts for only 2% of total ethylene oxide use. In Belgium, an estimated 0.07% of the total consumption of ethylene oxide was used in sterilization operations in health care and medical products industries (Wolfs et al., 1983).
The type of equipment and operation of sterilization facilities can influence the quantity of ethylene oxide released. Essentially, all facilities that control sterilizer vent emissions (usually greater than 99% removal efficiency) have recirculating-water vacuum pumps, which result in no loss of ethylene oxide through the water drain (Meiners and Nicholson, 1988; U.S. EPA, 1992, 1994). The Standards Council of Canada recommends that a liquid/gas separator be installed so that the liquid effluent is directed to the sanitary floor drain and the gas effluent is discharged through the local exhaust ventilation (CSA, 1991). In those facilities that use once-through water-sealed vacuum pumps, some ethylene oxide dissolved in the water will be directed to a floor drain and will likely volatilize to the atmosphere at an outdoor ground-level drain near the facility or wastewater treatment facility (U.S. EPA, 1992; WCB, 1994). Hospital sterilizer discharge experiments conducted in the United States on catalytic oxidation systems revealed absorption of ethylene oxide between 22 and 78% in the wastewater (Leclair et al., 1988; Meiners and Nicholson, 1988). Examination of the vacuum pump water discharge revealed rapid and complete loss of the ethylene oxide within 7 minutes from an initial concentration of 15 000 mg/L (Meiners and Nicholson, 1988).
In summary, the available data from point sources of release of ethylene oxide indicate that essentially all releases are to the atmosphere and that there has been a notable reduction in emissions in Canada, as evidenced by release trends reported through NPRI, progress through the ARET program and available information on releases from sterilization facilities. It is not clear how the new Shell Chemicals plant in Scotford, Alberta, will impact on the release trend; however, the predicted increase in production volume could slow this decline.
2.3 Exposure characterization
2.3.1 Environmental fate
Based on empirical fate data, release of ethylene oxide to the atmosphere is unlikely to result in transfer to other environmental compartments in significant quantities. Reaction half-life in the atmosphere may be significantly long (t1/2 between 38 and 382 days); however, evidence also indicates that washout by precipitation can be important where subsequent hydrolysis can occur (t 1/2 between 9 and 14 days). Revolatilization from the water is also expected to be rapid (t 1/2 ~1 hour). On the basis of a low log Kow (-0.30), the potential for bioaccumulation of ethylene oxide is expected to be very low. As a result of its high water solubility and vapour pressure, ethylene oxide is not expected to bioaccumulate or accumulate in sediment or soil.
2.3.1.1 Air
The atmospheric half-lives for ethylene oxide following vapour-phase reactions with photochemically produced hydroxyl radicals, assuming an atmospheric concentration of hydroxyl radicals of 1 x 106 radicals/cm3, were estimated at 120 days (Atkinson, 1986), 99 days (Lorenz and Zellner, 1984), 151 days (Zetzsch, 1985) and between 38 and 382 days (Howard et al., 1991).
If one considers that the hydroxyl radical concentration changes as a function of light duration and intensity, then the half-life can vary with latitude across Canada. By similar calculation, using an estimated hydroxyl radical concentration for Montréal, Quebec, around the March equinox of 8.6 x 105 radicals/cm3 (Bunce, 1997) and the experimental rate constant of 0.053 x 10-12 cm3/molecule per second (Zetsch, 1985), a half-life of 176 days is estimated for that location.
The theoretical atmospheric lifetimes (approximately 1.43 x t1/2) for ethylene oxide were estimated at ~200 days (Bunce, 1996) and 330 days (Winer et al., 1987) and were calculated based on the reaction with hydroxyl radicals at a concentration of 8.0 x 105 and 1.0 x 106 radicals/cm3, respectively. Such lifetimes are expected to be long enough to allow a very small percentage of the amount emitted to reach the stratosphere (Bunce, 1996).
Ethylene oxide has a very high water solubility (completely miscible), which would suggest that some washout via precipitation can be expected; however, its high vapour pressure (~146 kPa) and rapid volatilization rate may limit the effectiveness of this process. An examination of the effect of atmospheric precipitation was conducted in a laboratory setting (Winer et al., 1987), resulting in evidence that washout has little impact on reducing atmospheric concentrations.
2.3.1.2 Water
Ethylene oxide is expected to undergo numerous fate processes in water, including evaporation, hydrolysis, and aerobic and anaerobic biodegradation. Evaporation from water appears to be a significant removal process. The reported experimental aquatic half-life for evaporation of ethylene oxide in water is 1 hour with no wind and 0.8 hours with a 5 m/s wind (Conway et al., 1983). Ethylene oxide degrades in water by hydrolysis and other nucleophilic reactions (U.S. EPA, 1985). Ethylene oxide is hydrolyzed in fresh water to ethylene glycol and in salt water to ethylene glycol and ethylene chlorohydrin. The half-life was estimated experimentally to be 12-14 days for hydrolysis at pH 5-7 in fresh water and 9-11 days in salt water (Conway et al., 1983). The aqueous aerobic biodegradation half-life of ethylene oxide was approximately 20 days from a lightly seeded biological oxygen demand (BOD) test, and the rate in a biological waste treatment system is expected to be much faster (Conway et al., 1983). Based on the BOD test results of Bridié et al. (1979a) and Conway et al. (1983), Howard et al. (1991) estimated the unacclimated aqueous biodegradation half-life to be from 1 to 6 months. The aqueous anaerobic half-life, based on the estimated aerobic biodegradation half-life, is 4-24 months (Howard et al., 1991). The 5-day BOD was 3% of the theoretical oxygen demand (ThOD) of 1.82 g/g (Bridié et al., 1979a).
2.3.1.3 Soil and sediment
Ethylene oxide is miscible in water and poorly adsorbed to soil, so it has the potential to leach into groundwater (HSDB, 1999). Because of its high vapour pressure (146 kPa), a spill of ethylene oxide to soil will result in most volatilizing to the atmosphere, with only a small fraction infiltrating the soil. Evaporation will continue within the soil, but at a reduced rate (Environment Canada, 1985). Dilution with water will reduce the velocity at which the ethylene oxide moves downward and at the same time diminish the vapour pressure and reduce the rate of evaporation. Upon reaching the groundwater table, ethylene oxide will move in the direction of groundwater flow. The half-lives for hydrolysis in groundwater and soil are estimated to be between 10.5 and 11.9 days, based on measured rate constants at pH 5, 7 and 9 (Mabey and Mill, 1978; Howard et al., 1991). In general, volatilization is the primary removal mechanism, but ethylene oxide is expected to hydrolyze and be biodegraded relatively rapidly in most soils.
There is no information available on the environmental fate of ethylene oxide in sediment. Because of its physical and chemical properties, ethylene oxide is not expected to be sorbed by sediment or soil.
2.3.1.4 Biota
There are no reported levels of ethylene oxide in environmental biota. Ethylene oxide has a high vapour pressure (146 kPa) and high water solubility (infinitely soluble) (Table 1). On the basis of a low log Kow (-0.30), the potential for bioaccumulation of ethylene oxide is expected to be very low (Verschueren, 1983; Howard, 1989).
2.3.1.5 Environmental partitioning
Fugacity modelling was conducted to characterize key reaction, intercompartment and advection (movement out of a system) pathways for ethylene oxide and its overall distribution in the environment. A steady-state, non-equilibrium model (Level III fugacity model) was run using the methods developed by Mackay (1991) and Mackay and Paterson (1991). All physical and chemical property input values were selected from a compilation of literature values based on criteria for integrity (see DMER and AEL, 1996, for details). The key parameters are shown in Table 2.
Based on the ChemCAN Level III fugacity model, which depicts a mixedwood plain region of a densely populated region of southern Ontario, it is estimated that ethylene oxide will have an overall persistence of 3 days in that region from a reaction persistence estimated at 70 days. Owing to its short overall persistence, higher concentrations will likely be centralized in areas close to discharges. Based on the 1993 NPRI release volume of 53 200 kg to the atmosphere in southern Ontario (NPRI, 1993), the average steady-state levels in the southern Ontario region are estimated at 1.02 ng/m3 in air (344 kg), 0.067 ng/L in water (99.0 kg), 6.03 x 10-5 ng/g in soil (0.858 kg) and 3.27 x 10-5 ng/g in sediment (0.034 kg). Bioaccumulation is not expected (DMER and AEL, 1996).
Note: To ensure conservative estimates, some half-life values used for modelling were longer than actual estimated values.
The concentrations of ethylene oxide predicted above are based on the assumption that air entering southern Ontario from neighbouring regions contains no ethylene oxide. Estimates of concentrations of ethylene oxide in air in the 48 contiguous states of the United States, derived from atmospheric dispersion modelling and U.S. emission inventories, are available from the U.S. EPA's Cumulative Exposure Project (Woodruff et al., 1998). Mean concentrations predicted for 1990 in Michigan and New York, which border southern Ontario, were 4.9 ng/m3 and 5.9 ng/m3, respectively. When the average of these concentrations was assumed for the concentration of ethylene oxide in air advected into southern Ontario, the concentrations predicted by the ChemCAN model increased approximately 6-fold, to 6.2 ng/m3 in air, 0.4 ng/L in water, 3.7 x 10-4 ng/g in soil and 2.0 x 10-4 ng/g in sediment. Concentrations predicted in the ChemCAN auxiliary compartments of terrestrial animals and terrestrial plants were 4.3 x 10-5 ng/g and 1.4 x 10-3 ng/g, respectively, when this additional advective input was included in the fugacity modelling (Health Canada, 1999a).
2.3.2 Environmental concentrations
2.3.2.1 Ambient air
Data on concentrations of ethylene oxide in ambient air in Canada are very limited.
Information on levels in emissions from Canadian production, processing or sterilization facilities was not identified.
Ethylene oxide was detected at concentrations of 3.7, 3.9 and 4.9 µg/m3 in 3 of 50 24-hour samples of air collected outside of randomly selected residences during a multimedia exposure study conducted in Canada (Health Canada, 1999a). The censored mean value was 0.34 µg/m3 when a concentration equivalent to one-half the limit of detection (i.e., ½ x 0.19 µg/m3 = 0.095 µg/m3) was assumed for the 47 samples in which ethylene oxide was not detected. Ethylene oxide was detected at 3 (or 33%) of 9 locations in Alberta, but at none of the 35 locations in Ontario or the 6 locations in Nova Scotia during this study (Health Canada, 1999a).
Based on data on 1993 air quality modelling predictions from a Canadian production facility, obtained from a survey of Canadian industry carried out under the authority of Section 16 of CEPA (Environment Canada, 1997b), it was estimated that 1-hour average ground-level concentrations of ethylene oxide would exceed 12 µg/m3 a total of 17 hours a year in the immediate vicinity of the plant. Predicted maximum 1-hour-average ground-level concentrations ranged from 3.7 to 20.1 µg/m3 at distances of 5 and 2.7 km from the plant, respectively. No measurements were available to validate these predictions.
Notes:
- Momentum and buoyancy effects are not considered.
- The model estimated a 24-hour average ethylene oxide concentration in ppb.
Estimated maximum average daily concentrations of ethylene oxide in the vicinity of Canadian hospitals were 0.26, 0.83, 1.3 and 2.12 µg/m3 between 100 and 70 m from the emission source and from stack heights of 30, 18, 15 and 12 m, respectively. Concentrations closer to or farther from the source are predicted to be less. The estimate was based on the U.S. EPA "SCREEN3" Gaussian plume model, which incorporates source-related and meteorological factors to estimate pollutant concentration from continuous sources. The model assumes that the pollutant does not undergo any chemical reactions and that no other removal processes, such as wet or dry deposition, act on the plume during its transport from the source (U.S. EPA, 1995) (see Table 3 for input parameters).
In an assessment of emissions and concentrations of ethylene oxide throughout California, mean 24-hour ambient air concentrations sampled in Los Angeles ranged from 0.038 to 955.7 µg/m3 (n = 128) (Havlicek et al., 1992). The authors reported that heavy usage of ethylene oxide within the Los Angeles basin coupled with restricted airflow out of the basin likely led to the large range of ambient air concentrations. Air concentrations sampled in northern California ranged from 0.032 to 0.40 µg/m3 (n = 36). At remote coastal locations in California, ethylene oxide concentrations ranged from 0.029 to 0.36 µg/m3 (n = 22). The authors warn that it is difficult to make any definitive statements regarding the spatial and temporal distribution of ethylene oxide based on the samples collected. Air concentrations were found to be very variable, especially in urban areas, with 100-fold shifts in air concentrations over a few minutes. There was a high degree of local variability that would be consistent with the release of ethylene oxide during a sterilization cycle.
In a modelling study on point sources of ethylene oxide from sterilizers, information on characteristics of 17 sources in Duval County, Florida, and their emissions was collected in survey questionnaires in 1989 (Tutt and Tilley, 1993). Peak short-term and long-term ambient concentrations of ethylene oxide as a result of emissions from four sterilization facilities identified were estimated based on the U.S. EPA's SCREEN and Industrial Source Complex Short-Term dispersion models. These included a commercial spice fumigation facility (with estimated annual emissions of ethylene oxide of 1959.5 kg per year) and three hospitals of decreasing emission profiles (i.e., from 210.9 to 2.1 kg per year). The predicted maximum average annual concentrations from the two highest emitters were 11 µg/m3 and 2 µg/m3, both at distances 32 m from their respective point sources.
2.3.2.2 Indoor air
Ethylene oxide was detected at a concentration of 4 µg/m3 in only 1 of 50 24-hour samples of air collected inside randomly selected residences during a multimedia exposure study conducted in Canada (Health Canada, 1999a). The censored mean value was 0.17 µg/m3 when a concentration equivalent to one-half the limit of detection (i.e., ½ x 0.19 µg/m3 = 0.095 µg/m3) was assumed for the 49 samples in which ethylene oxide was not detected. Ethylene oxide was detected at concentrations of 5 µg/m3 in 3 of 24 personal air samples collected from an occupant of each of the 50 residences (Conor Pacific Environmental, 1998).
2.3.2.3 Drinking water
Data on concentrations of ethylene oxide in drinking water were not identified.
2.3.2.4 Surface water
Data on concentrations of ethylene oxide in surface water in Canada were not identified.
2.3.2.5 Sediment and soil
Data on concentrations of ethylene oxide in sediments and soils in Canada were not identified.
2.3.2.6 Biota
Data on concentrations of ethylene oxide in biota in Canada were not identified.
2.3.2.7 Food
Data on the levels of ethylene oxide in foodstuffs consumed in Canada were not identified.
Ethylene oxide was detected in 96 (or 47%) of 204 samples of food products taken from retail stores in Denmark in 1985 (Jensen, 1988). The reported concentrations reflect the total amount of ethylene chlorohydrin and ethylene oxide present at the time of analysis. These concentrations ranged from <0.05 to 1800 mg/kg (or µg/g) in the individual samples without correction for recoveries. Ethylene oxide was detected frequently among 24 samples of spices (Jensen, 1988) at a mean concentration of 84 µg/g and a maximum concentration of 580 µg/g.
Ethylene oxide was detected, but not quantified, in 1 of 2372 samples of eggs and in 1 of 3262 samples of fish collected in the United States in 1975 as part of the Food and Drug Administration Monitoring Program (1970-1976) (Duggan et al., 1983).
2.3.2.8 Consumer products
Ethylene oxide may be present in tobacco as a result of its use as a fumigant and sterilizing agent (ATSDR, 1990). It has been detected in smoke from fumigated and unfumigated tobacco at levels of 0.3 and 0.02 µg/mL, respectively (Binder, 1974).
Ethylene oxide may also be present as a contaminant of skin care products. Current commercial preparations of polyglycol ethers may contain residues of ethylene oxide monomer up to approximately 1 µg/g, according to a European study (Filser et al., 1994). Kreuzer (1992) reported concentrations of ethylene oxide monomer in skin care products ranging from 1.9 to 34 nmol/cm3 (0.08-1.5 mg/L) and a range of maximum skin penetration of ethylene oxide of 1.0-14% in various product formulations.
2.3.2.9 Medical devices
Ethylene oxide is the most common agent currently used for sterilizing disposable dialysers, blood tubing and heat-sensitive medical items and has almost completely replaced formalin for this purpose (Henne et al., 1984; Babich, 1985). Ethylene oxide may be absorbed by medical equipment during sterilization and may remain there as the unchanged compound or as one of its reaction products (WHO, 1985). In studies conducted outside of Canada, concentrations of residual ethylene oxide in medical devices immediately following their sterilization have ranged up to 1 or 2% (Gillespie et al., 1979; Gilding et al., 1980). These concentrations generally declined rapidly after a few days' aeration, although levels exceeding 100 ppm (183 mg/m3) were sometimes measured following aeration.
2.4 Effects characterization
2.4.1 Ecotoxicology
Information on the toxicity of ethylene oxide to natural aquatic and terrestrial organisms is limited. A brief summary of effects is presented below, with an emphasis on the most sensitive endpoints. A large number of studies have shown that ethylene oxide can induce dose-dependent genetic mutations in various types of biota, including plants, fungi, insects, mammalian cell cultures and bacteria (U.S. EPA, 1985; WHO, 1985; Dellarco et al., 1990; IARC, 1994; BUA, 1995; see also Section 2.4.3.4). The actual population-level impact to wildlife from mutagenic endpoints is not clear. In general, however, these effects occur at levels of ethylene oxide exposure similar to or slightly lower than those observed to induce other effects. A more extensive description of the environmental effects is provided in Environment Canada (1999).
2.4.1.1 Aquatic organisms
Toxicity of ethylene oxide to bacterial cultures has been examined primarily in terms of mutagenicity (Dellarco et al., 1990). Conway et al. (1983) examined direct toxicity by determining the IC50 for the effect of ethylene oxide on activated sludge microorganisms in a 16-hour bacterial toxicity test at 22°C. The IC50 was in the range of 10-100 mg/L (Conway et al., 1983).
In a modified Ames test, direct increases in revertant bacterial mutations in Salmonella typhimurium strains TA1535 and TA100 were observed (Pfeiffer and Dunkelberg, 1980). Similarly, a dose-response relationship was observed for mutation induction in Escherichia coli Sd-4 (Hussain, 1984). There was a linear relationship for induction of guanine alkylation at concentrations between 2.6 and ~1000 mg/L for 1-hour exposures. Bacterial survival remained essentially constant at 100% at all dose levels (Hussain, 1984).
Examination of point mutation studies conducted using various strains of bacteria, including DNA repair-deficient strains, revealed positive dose-response relationships following exposure in liquid suspensions up to 4210 mg/L (Dellarco et al., 1990). From the above studies and others, it is difficult to interpret the effects of ethylene oxide on natural bacterial populations in the laboratory when one considers bacterial strains, reproductive potential, natural variability, DNA repair mechanisms and population resilience.
Ethylene oxide appears to be slightly less toxic to invertebrates than to microorganisms. Conway et al. (1983) performed a U.S. EPA standard static acute toxicity test on Daphnia magna that yielded 24-hour LC50 values of 260-300 mg/L and 48-hour LC50 values of 137-300 mg/L. In acute toxicity tests on brine shrimp (Artemia salinia) under similar conditions, 24-hour LC50s were between 350 and 500 mg/L (n = 3), and 48-hour LC50 values were between 490 and 1000 mg/L (n = 3) (Conway et al., 1983).
Fish are moderately sensitive to ethylene oxide. Bridié et al. (1979b) examined the acute toxicity of ethylene oxide to goldfish (Carassius auratus) and reported a 24-hour LC50 of 90 mg/L at 20°C. In acute static toxicity tests performed according to U.S. EPA standards using fathead minnows (Pimephales promelas) under aerated conditions, under sealed oxygen or under no aeration, 24-hour LC50s were 274, 86 and 90 mg/L, respectively. With no aeration, the 48- and 96-hour LC50s were 89 and 84 mg/L, respectively (Conway et al., 1983).
2.4.1.1.1 Toxicity of breakdown products
Ethylene glycol and ethylene chlorohydrin are the principal breakdown products of ethylene oxide in water. In acute toxicity tests with ethylene glycol, 24-hour LC50s were >10 000, >10 000 and >20 000 mg/L for the fathead minnow, Daphnia magna and brine shrimp, respectively for ethylene chlorohydrin, 24-hour LC50s were 768, 675 and >1000 mg/L for the fathead minnow, D. magna and brine shrimp, respectively (Conway et al., 1983).
2.4.1.2 Terrestrial organisms
Ethylene oxide produces gene mutations in plant cells, including barley, rice and peas, exposed in vitro (Ehrenberg et al., 1956, 1959; Blixt et al., 1963; Shulovská et al., 1969; Jana and Roy, 1975; Migliore et al., 1982). Chromosome damage and sister chromatid exchange were observed in barley, wheat and spiderwort (Tradescantia paludosa) pollen (Smith and Lotfy, 1954; Ehrenberg et al., 1956, 1959; Mackey, 1968; Moutschen-Dahmen et al., 1968). Ehrenberg et al. (1956) reported a 5-fold increase in sterility caused by chromosomal aberrations in barley seeds treated at a gaseous ethylene oxide concentration of 1.5 x 106 mg/m3 (80%) for 6 days. Second-generation chlorophyll gene mutations increased 33 times over controls under this treatment. Barley seeds soaked for 2 hours in solutions of 3084 and 11 894 mg/L induced second-generation chlorophyll gene mutations 3.7- and 13.8-fold over controls. Jana and Roy (1975) determined that for two genotypes of rice (Oryza satica), mutagenic efficiency decreased with increasing concentration of ethylene oxide. Concentrations ranged from 888 to 6167 mg/L, and exposure was for 8 hours.
In an examination of the control of pathogenic fungi (Fusarium, Alternaria and Helminthosporium spp.) on sorghum (Sorghum vulgare Pers.) through exposure to ethylene oxide, ethylene oxide applied to a filter paper disk at 8 mg/L was 92.3% effective in controlling fungal growth and 100% effective in inhibiting the viability of the sorghum seed (Raghunathan et al., 1969). It should be noted that interpretation of this study is difficult because concentrations applied to the filter paper disks are not easily related to air or soil concentrations.
Although data are limited, insects appear to be relatively insensitive to atmospheric exposure to ethylene oxide. In 24-hour fumigation experiments, there was an increase in the control-corrected mortality rates from 24.5 to 98.6% in the khapra beetle (Trogoderma granarium) as the concentration of ethylene oxide increased from 1000 to 3000 mg/m3, respectively. Reproduction of the surviving beetles did not differ significantly from that of untreated controls (Rajendran, 1982). Rajendran and Shivaramaiah (1985) studied the effect of 24-hour exposure to ethylene oxide at concentrations ranging from 250 to 1500 mg/m3 on the reproductive rate of the lesser grain borer (Rhyzopertha dominica F.). Only concentrations above 500 mg/m3 had any significant effect (p = 0.01) on reproductive rate. Chromosome damage has been observed in insects exposed to ethylene oxide, including observations of gene mutations in Drosophila melanogaster from a sex-linked recessive lethal test and in vitro evidence of chromosomal breaks and translocations (WHO, 1985).
No information is available describing the effects of ethylene oxide on birds or wild mammals. Laboratory animals are therefore used as surrogates for wildlife. The chronic reproductive effects in rats following inhalation of ethylene oxide at 183 mg/m3, reported by Snellings et al. (1982b), are assumed to represent effects on wild rodent species and are selected as the most environmentally significant measurement endpoint for the assessment of the effects of ethylene oxide on the terrestrial environment (see also Section 2.4.3.5.1).
2.4.2 Abiotic atmospheric effects
Worst-case calculations to determine if ethylene oxide has the potential to contribute to the depletion of stratospheric ozone, ground-level ozone formation or climate change are presented below (Bunce, 1996).
The Ozone Depletion Potential (ODP) is 0, since ethylene oxide is not a halogenated compound (Bunce, 1996). The Global Warming Potential (GWP) was calculated to be 0.031 (relative to the reference compound CFC-11, which has a GWP of 1), based on the following formula (Bunce, 1996):
GWP = (tethylene oxide /tCFC-11) x (MCFC-11 /Methylene oxide) x
(Sethylene oxide /SCFC-11)
where:
- tethylene oxide is the lifetime of ethylene oxide
(0.60 years), - tCFC-11 is the lifetime of CFC-11 (60 years),
- MCFC-11 is the molecular weight of CFC-11
(137.5 g/mol), - Methylene oxide is the molecular weight of ethylene
oxide (44.05 g/mol), - Sethylene oxide is the infrared absorption strength
of ethylene oxide (2389/cm2 per atmosphere, default), and - SCFC-11 is the infrared absorption strength of
CFC-11 (2389/cm2 per atmosphere).
The Photochemical Ozone Creation Potential (POCP) was estimated to be 0.5 (relative to the value of an equal mass of the reference compound ethene, which has a POCP of 100), based on the following formula (Bunce, 1996):
POCP = (kethylene oxide /Kethene) x (Methene /Methylene oxide) x 100
where:
- kethylene oxide is the rate constant for the reaction
of ethylene oxide with OH radicals (7.0 x 10-14 cm3/mol per second), - Kethene is the rate constant for the reaction of ethene with OH radicals (8.5 x 10-12 cm3/mol per second),
- Methene is the molecular weight of ethene (28 g/mol), and
- Methylene oxide is the molecular weight of ethylene oxide (44.05 g/mol).
These figures suggest that the potential contribution of ethylene oxide to stratospheric ozone depletion and ground-level ozone creation is negligible. The potential to contribute to climate change is also considered to be minimal (Bunce, 1996).
2.4.3 Experimental animals and in vitro
2.4.3.1 Acute toxicity
Ethylene oxide is of low acute toxicity following inhalation, with 4-hour LC50s of 1460, 835 and 960 ppm (2672, 1528 and 1757 mg/m3) for rats, mice and dogs, respectively, being reported by Jacobson et al. (1956). Mortality of 80-100% was observed in male and female mice exposed to 800 ppm (1464 mg/m3) ethylene oxide for 4 hours; no mortality was observed at the next lowest concentration (400 ppm [732 mg/m3]) (NTP, 1987). LD50s for ethylene oxide administered orally (in water) were 330 mg/kg-bw for male rats and 280 and 365 mg/kg-bw for female and male mice, respectively (Smyth et al., 1941; Woodard and Woodard, 1971).
The lungs and nervous system are the principal organs affected following exposure to acutely toxic levels of ethylene oxide.
2.4.3.2 Short-term and subchronic toxicity
Available data on the repeated-dose toxicity of ethylene oxide are limited, being restricted primarily to inhalation studies in which animals were exposed to single concentrations.
Increased mortality was observed following the inhalation exposure of rats, mice, guinea pigs, rabbits and monkeys to concentrations of ethylene oxide ranging from 732 to 1500 mg/m3 for 10 days to 8 weeks (Hollingsworth et al., 1956; Jacobson et al., 1956; Snellings, 1982; NTP, 1987). The exposure of rats to ethylene oxide at concentrations between 180 and 915 mg/m3 for several weeks produced hematological effects (including decreases in hemoglobin and red blood cells and a reduction in the number of lymphocytes), changes in clinical chemistry, as well as histopathological alterations in various tissues, including the nasal mucosa, thymus and testes (Jacobson et al., 1956; Snellings, 1982; Mori et al., 1990). Effects observed in mice exposed to ethylene oxide at 810 mg/m3 for 6 hours per day, 5 days per week, for 3 weeks included reduced body weight gain, poor coordination of the hind quarters, irregular breathing, convulsions and red urine. In both rats and mice, body weight gain was reduced following repeated exposure to levels of ethylene oxide as low as 90 mg/m3 (6 hours per day, 5 days per week, for approximately 7 weeks) (Snellings, 1982).
Common effects observed following the subchronic exposure of rats to ethylene oxide have been hematological and metabolic disturbances. Reductions in hemoglobin concentration, hematocrit and red blood cell counts, accompanied by increases in reticulocytes, were observed in rats exposed to 915 mg ethylene oxide/m3 for 6 hours per day, 3 times per week, for 13 weeks (Fujishiro et al., 1990; Mori et al., 1990). This exposure regimen also produced declines in the activities of glutathione reductase and creatine kinase in blood and various tissues (Katoh et al., 1988, 1989; Matsuoka et al., 1990; Mori et al., 1990; Fujishiro et al., 1991), as well as increased hepatic lipid peroxidation (Katoh et al., 1988, 1989). Other effects observed in rats following subchronic exposure to ethylene oxide at concentrations ranging from 370 to 915 mg/m3 included those on the nervous system (ataxia and axonal degeneration in the hind limbs and spinal cord) (Hollingsworth et al., 1956; Ohnishi et al., 1985, 1986; Matsuoka et al., 1990; Mori et al., 1990), disturbances in hepatic porphyrin-heme metabolism (Fujishiro et al., 1990) and histopathological changes in the testes, kidneys and lungs (Hollingsworth et al., 1956). Effects observed in mice exposed to ethylene oxide are similar to those in rats. Hematological changes, including decreases in red blood cell count, hemoglobin concentration, hematocrit, packed cell count, bone marrow cellularity and numbers of lymphocytes, have been observed in mice exposed (6 hours per day, 5 days per week, for 10 or 11 weeks) to ethylene oxide at concentrations of 425 or 467 mg/m3 (Snellings et al., 1984a; Popp et al., 1986). Thymic lymphocytic necrosis, renal tubular necrosis and lymphocytic necrosis of the spleen have been reported after the exposure of mice to much higher concentrations (i.e., 1098 mg/m3) for 6 hours per day, 5 days per week, for approximately 13 weeks; renal tubular degeneration was observed at concentrations as low as 183 mg/m3 (NTP, 1987). Exposure to concentrations as low as 86 mg/m3 reduced locomotor activity (Snellings et al., 1984a).
In a limited, poorly reported study involving the exposure of rats, mice, rabbits, guinea pigs and monkeys to concentrations of ethylene oxide ranging from 90 to 640 mg/m3, a reduction in growth was evident in all species at the highest concentration (statistical significance not reported) (Hollingsworth et al., 1956). Atrophy of the hind leg muscles was observed in rabbits and monkeys after exposure to ≥370 mg ethylene oxide/m3, for exposure periods ranging from 7 to 32 weeks (exact exposure periods not clearly specified) (Hollingsworth et al., 1956).
Guinea pigs exposed to 640 mg ethylene oxide/m3 exhibited degeneration of the tubules of the testes, with "replacement fibrosis" as well as slight fatty degeneration in the adrenal cortex in females (Hollingsworth et al., 1956). In another limited study, decreases in red blood cell count, hemoglobin and hematocrit were observed in two of three dogs exposed to 183 mg ethylene oxide/m3 for 6 months (Jacobson et al., 1956). No differences in hematological parameters (red or white blood cell counts, hematocrit, hemoglobin or white cell differential) were observed in rabbits exposed to 458 mg ethylene oxide/m3 for 12 weeks, compared with unexposed controls (Yager and Benz, 1982).
In the only short-term study identified on the oral toxicity of ethylene oxide, there was a loss of body weight, gastric irritation and slight liver damage following exposure of rats to 100 mg ethylene oxide/kg-bw, 5 times per week, for a total of 15 doses in 21 days (Hollingsworth et al., 1956). Information on the subchronic toxicity of ethylene oxide following ingestion was not identified.
2.4.3.3 Chronic toxicity and carcinogenicity
2.4.3.3.1 Chronic toxicity
Non-neoplastic effects associated with chronic exposure to ethylene oxide have not been investigated extensively, most studies having focussed on the carcinogenicity of this substance. In several investigations conducted with rats exposed to ethylene oxide for 2 years, significant reductions in body weight gain at concentrations as low as 60.4 mg/m3 and decreased survival time at exposures of ≥92 mg/m3 have been observed (Lynch et al., 1984a,b; Snellings et al., 1984b; Garman et al., 1985; Garman and Snellings, 1986). Additional non-neoplastic effects observed at exposures of ≥92 mg ethylene oxide/m3 include increased levels of aspartate aminotransferase in serum, reduced absolute kidney and adrenal weights, an increased incidence of inflammatory lesions in the lungs, nasal cavity, trachea and internal ear, proliferative and degenerative lesions in the adrenal cortex, as well as an increased incidence of multifocal mineralization of the posterior layers of the choroid/sclera portion of the eye (Lynch et al., 1984a,b). Skeletal muscular atrophy (in the absence of sciatic nerve neuropathy) was noted following exposure to 183 mg ethylene oxide/m3 (Lynch et al., 1984a,b).
No exposure-related effects upon survival, body weight gain, clinical signs or other non-neoplastic endpoints (examined in a wide range of tissues) were observed in B6C3F1 mice exposed to 92 or 183 mg ethylene oxide/m3 for 2 years (NTP, 1987). (Notably, in subchronic studies with this strain of mice, neuromuscular screening tests revealed effects on locomotor activity at a concentration of 86 mg ethylene oxide/m3 [Snellings et al., 1984a].)
Monkeys exposed for 2 years to ≥92 mg ethylene oxide/m3 developed axonal dystrophy in the nucleus gracilis of the medulla oblongata of the brain, along with demyelination in the distal portion of the fasciculus gracilis (Sprinz et al., 1982; Lynch et al., 1984b). Weight gain was significantly reduced following exposure to 183 mg ethylene oxide/m3 (Lynch et al., 1984a,b; Setzer et al., 1996).
2.4.3.3.2 Carcinogenicity
Substance-related increases in a variety of tumour types have been observed in rodents exposed to ethylene oxide. In two studies, inhalation exposure increased the incidence of mononuclear cell leukemia1 and gliomas of the brain in F344 rats of both sexes and of peritoneal mesotheliomas in male rats. In mice, increased incidences of alveolar/bronchiolar adenomas or carcinomas and Harderian gland papillary cystadenomas were observed in both sexes, while the incidences of malignant lymphomas, uterine and mammary gland adenocarcinomas and mammary gland adenocarcinomas or adenosquamous carcinomas (combined) were increased in females. An increase in squamous cell carcinomas of the forestomach was observed in female rats following the administration (by gavage) of ethylene oxide; the subcutaneous injection of ethylene oxide to female mice induced local fibrosarcomas.
In male (n = 80 per group) Fischer 344 rats exposed to 0, 50 or 100 ppm (0, 92 or 183 mg/m3) ethylene oxide for 7 hours per day, 5 days per week, for 104 weeks, the incidence of mononuclear cell leukemia was 24/77, 38/79 (p = 0.03) and 30/76, respectively (Lynch et al., 1984a,b). Among animals in the control, low-and high-concentration groups, the incidence of peritoneal mesotheliomas and mixed cell gliomas in brain tissue was 3/78, 9/79 and 21/79 (p = 0.01) and 0/76, 2/77 and 5/79 (p < 0.05), respectively. There was a significant concentration-related trend between the incidence of mesotheliomas and ethylene oxide exposure.
Similar results were observed when groups (n = 120 per sex) of male and female Fischer 344 rats were exposed to 0, 10, 33 or 100 ppm (0, 18.3, 60.4 or 183 mg/m3) ethylene oxide for 6 hours per day, 5 days per week, for up to 2 years (Snellings et al., 1984b; Garman et al., 1985; Garman and Snellings, 1986). The incidence of mononuclear cell leukemia (in animals sacrificed after exposure for 2 years) was (data from two control groups pooled) 13/97, 9/51, 12/39 and 9/30 (for males) and 11/116, 11/54, 14/48 and 15/26 (p < 0.001) (for females) in the control, low-, mid- and high-concentration groups, respectively. Trend analysis revealed a significant association for both sexes, although the increase was clearly concentration related only in females and was significantly different from the control group in females at the highest concentration only (i.e., 183 mg/m3) (Snellings et al., 1984b). Among males exposed to 0, 18.3, 60.4 or 183 mg ethylene oxide/m3, the incidence of peritoneal mesotheliomas (in animals sacrificed after exposure for 2 years) was 2/97, 2/51, 4/39 and 4/30, respectively; trend analysis indicated a relationship between ethylene oxide exposure and tumour induction after adjustment for mortality (Snellings et al., 1984b). Concentration-related increases in primary brain tumours (gliomas, malignant reticuloses and granular cell tumours) were observed in both sexes of rats exposed to ethylene oxide (Garman et al., 1985; Garman and Snellings, 1986). Among animals in the control, low-, mid- and high-concentration groups, the incidence (expressed as number of animals with tumour/number alive at the time the first tumour was observed in any group) of such brain tumours (combined) was 1/181, 1/92, 5/85 (p = 0.027) and 7/87 (p = 0.004) (for males) and 1/188, 1/94, 3/92 and 4/80 (p = 0.058) (for females), respectively. The incidence of subcutaneous fibroma (15/58) was significantly increased in male rats in the highest exposure group (i.e., 183 mg/m3) (Snellings et al., 1984b). The increase in the incidence of mononuclear leukemia, mesothelioma and brain tumours in these animals occurred during the later stages of this study (i.e., after about 20-24 months of exposure to ethylene oxide) (Snellings et al., 1984b; Golberg, 1986).
The chronic inhalation exposure of mice to ethylene oxide increased the incidence of tumours at sites different from those observed in rats. In groups (n = 50 per sex) of male and female B6C3F1 mice exposed to 0, 50 or 100 ppm (0, 92 or 183 mg/m3) ethylene oxide for 6 hours per day, 5 days per week, for 102 weeks, there was a significant concentration-related increase in the incidence of alveolar/bronchiolar carcinomas (NTP, 1987). Among animals in the control, low-and high-concentration groups, the incidence of alveolar/bronchiolar carcinomas was 6/50, 10/50 and 16/50 (p = 0.019) (for males) and 0/49, 1/48 and 7/49 (p = 0.017) (for females), respectively. For males and females exposed to 0, 92 or 183 mg ethylene oxide/m3, the incidence of papillary cystadenoma within the Harderian gland was 1/43, 9/44 (p = 0.012) and 8/42 (p = 0.012) (males) and 1/46, 6/46 and 8/47 (p = 0.033) (females), respectively. In females, there were concentration-related increases in the incidence of malignant lymphomas of the hematopoietic system (9/49, 6/48 and 22/49 [p = 0.005]) and uterine adenocarcinoma (0/49, 1/47 and 5/49); the incidence of mammary adenocarcinoma and adenosquamous carcinoma (combined) was 1/49, 8/48 and 6/49, respectively (NTP, 1987).
In a study conducted with female A/J mice in which only the lungs were examined, a concentration-related increase in the incidence of pulmonary adenomas was observed following exposure (6 hours per day, 5 days per week) to 128 and 366 mg ethylene oxide/m3 for 6 months (Adkins et al., 1986).
In the only identified carcinogenicity study involving oral exposure, the intragastric administration of 7.5 or 30 mg ethylene oxide/kg-bw to female Sprague-Dawley rats twice weekly for 150 weeks produced a dose-related increase in the incidence of forestomach tumours (mainly squamous cell carcinomas) (Dunkelberg, 1982).
In female NMRI mice, the subcutaneous injection of ethylene oxide for 95 weeks (mean total doses up to 64.4 mg/mouse) yielded a significant dose-dependent increase in the number of tumours (i.e., sarcomas) at the site of injection (Dunkelberg, 1981). No skin tumours were observed in female ICR/Ha Swiss mice following the dermal application of approximately 100 mg ethylene oxide (10% in acetone) 3 times weekly for life (Van Duuren et al., 1965).
2.4.3.4 Genotoxicity
Ethylene oxide is a potent alkylating agent that has displayed genotoxic activity in almost all studies (reviewed in IARC, 1994). In in vitro testing, it induced DNA damage and gene mutations in bacteria, yeast and fungi and gene conversion in yeast. In mammalian cells, observed effects have included gene mutations, micronucleus formation, chromosomal aberrations, cell transformation, unscheduled DNA synthesis, sister chromatid exchange and DNA strand breaks. Notably, Hallier et al. (1993) observed that the frequency of sister chromatid exchange in human peripheral blood lymphocytes exposed in vitro to ethylene oxide was higher in cells isolated from individuals expressing low levels of theta-class glutathione S-transferase (GSTT1), compared with the effect in cells from subjects expressing higher levels of this enzyme.
The results of in vivo studies on the genotoxicity of ethylene oxide have also been consistently positive (see IARC, 1994) following ingestion, inhalation or injection. in vivo exposure to ethylene oxide induced gene mutation at the hypoxanthine phosphoribosyl transferase (Hprt) locus in mouse and rat splenic T-lymphocytes; sister chromatid exchange was induced in lymphocytes from rabbit, rat and monkey, in bone marrow cells from mouse and rat and in rat spleen. Increases in the frequency of gene mutations in the lung (lacI locus) (Sisk et al., 1997) and in T-lymphocytes (Hprt locus) (Walker et al., 1997a) have been observed in transgenic mice exposed to ethylene oxide via inhalation, at concentrations similar to those employed in carcinogenesis bioassays with this species (NTP, 1987).
In male Big BlueŽ (lacI transgenic) B6C3F1 mice exposed to 0, 50, 100 or 200 ppm (0, 92, 183 or 366 mg/m3) ethylene oxide for 6 hours per day, 5 days per week, for 4 weeks, the observed mean (ą SE) frequency of mutation at the Hprt locus in splenic T-lymphocytes was 2.2 (ą 0.03) x 10-6, 3.8 (ą 0.5) x 10-6 (p = 0.009), 6.8 (ą 0.9) x 10-6 (p = 0.001) and 14.1 (ą 1.1) x 10-6 (p < 0.001), respectively (Walker et al., 1997a). The frequency of Hprt mutations in splenic T-lymphocytes was reportedly increased (compared with unexposed controls) 5.0- to 5.6-fold in male F344 rats and (non-transgenic) male B6C3F1 mice exposed to 200 ppm (366 mg/m3) ethylene oxide for 6 hours per day, 5 days per week, for 4 weeks (Walker et al., 1997b). The mean (ą SD) frequency of lacI mutations in the lungs of male Big BlueŇ (lacI transgenic) B6C3F1 mice exposed to 0 or 200 ppm (0 or 366 mg/m3) ethylene oxide was 6.2 (ą 2.2) x 10-5 and 9.1 (ą 1.5) x 10-5 (p < 0.05), respectively (Sisk et al., 1997). The mean (ąSD) frequency of lacI mutations in the bone marrow, spleen and germ cells of these animals exposed to 0 or 200 ppm (0 or 366 mg/m3) ethylene oxide was 2.5 (ą 1.1) x 10-5 and 4.7 (ą 2.1) x 10-5, 4.2 (ą 0.7) x 10-5 and 5.4 (ą 3.2) x 10-5, and 3.4 (ą 2.3) x 10-5 and 2.9 (ą 1.1) x 10-5, respectively (Sisk et al., 1997). In a recent study (reported in abstract form), a 5-fold increase (compared with unexposed controls) in the frequency of lacI mutations in bone marrow cells from Big BlueŽ (lacI transgenic) B6C3F1 mice was observed following exposure to 200 ppm (366 mg/m3) ethylene oxide, 6 hours per day, 5 days per week, for 48 weeks; however, no increase was observed following exposures either to lower concentrations or for shorter periods (Recio et al., 1999).
In vivo exposure to ethylene oxide also induced heritable mutations or effects in germ cells in rodents (see IARC, 1994). Ethylene oxide induced dominant lethal effects in mice and rats and heritable translocations in mice. Dominant visible and electrophoretically detectable mutations were observed in the offspring of male mice exposed (by inhalation) to 200 ppm (366 mg/m3) ethylene oxide for 6 hours per day, 5 days per week, for 7 weeks and then mated. This exposure regimen ensured that all progeny originated from sperm exposed during the entire spermatogenic process (Lewis et al., 1986). In a study in which male (C3H x 101)F1 mice were exposed by inhalation to 0, 165, 204, 250 or 300 ppm (0, 302, 373, 458, or 549 mg/m3) ethylene oxide, 6 hours per day, 5 days per week, for 6 weeks, then daily for an additional 2.5 weeks, and subsequently mated to T-stock (or [SEC x 101]F1) females, the percent dominant lethals (p < 0.01 at concentrations ≥373 mg/m3, compared with controls) was 0 (0), 6 (8), 14 (13), 23 (24) and 60 (45), respectively (Generoso et al., 1990). The frequency of translocation carriers (p < 0.01 at all concentrations, compared with controls) among the progeny of these groups of ethylene oxide-exposed male mice mated to T-stock (or [SEC x C57BL]F1) females (data combined) was 1/2068 (0.05%), 32/1143 (2.8%), 52/1021 (5.1%), 88/812 (10.8%) and 109/427 (25.5%), respectively (Generoso et al., 1990).
2.4.3.5 Reproductive and developmental toxicity
2.4.3.5.1 Effects on reproduction
Degeneration of the seminiferous tubules and germ cells, decreased epididymal weight, decreased sperm count and an increase in the percentage of abnormal sperm were observed in Wistar rats exposed to ≥458 mg ethylene oxide/m3 for 13 weeks (Mori et al., 1989, 1991). When abnormal sperm heads were classified into immature and teratic types, the frequency of teratic types was increased at exposures of ≥92 mg/m3, although it was not concentration dependent (Mori et al., 1991). Decreased relative testicular weight was observed in rats after exposure to 915 mg ethylene oxide/m3 (Mori et al., 1989). In a limited study in rats, slight degeneration of the tubules in the testes was observed after exposure to 370 mg ethylene oxide/m3 for 25-32 weeks (Hollingsworth et al., 1956). Embryotoxic and fetotoxic effects have been observed in reproductive studies with rats after exposure of the dams via inhalation to concentrations of ethylene oxide between 183 and 275 mg/m3, prior to mating and throughout gestation. These effects included a decrease in the number of implantation sites per pregnant female, an increase in the incidence of resorptions, a decrease in the median number of pups born on day 0 postpartum per litter, as well as a lower ratio of the number of fetuses born to the number of implantation sites per female (Hackett et al., 1982; Snellings et al., 1982a,b; Hardin et al., 1983). Under these exposure conditions, adverse effects on the dams were not observed (based simply upon clinical appearance and demeanor).
Reproductive effects in mice are similar to those observed in rats. After exposure of female hybrid mice to 300 or 1200 ppm (549 or 2196 mg/m3) ethylene oxide for a period prior to mating, an increase in the number of resorption bodies and reductions in the number of implants per female and in the number of living embryos per female were observed (Generoso et al., 1987). Exposure of Swiss-Webster mice to 366 mg ethylene oxide/m3 for 5 days produced a concentration-related increase in the percentage of abnormal sperm (Ribeiro et al., 1987). A slight (7%) decline in absolute but not relative testicular weight was observed in B6C3F1 mice after exposure to 86 mg ethylene oxide/m3 for 10 weeks; however, no histological changes were observed in this tissue (Snellings et al., 1984a).
A decline in sperm count and motility was observed in monkeys exposed to concentrations of ethylene oxide as low as 92 mg/m3 for 24 months (Lynch et al., 1984b,c).
2.4.3.5.2 Developmental toxicity
Exposure of Sprague-Dawley rats to a maternally toxic concentration of 275 mg ethylene oxide/m3 either prior to mating and throughout gestation or only during various stages of gestation resulted in reduced fetal body weight and crown-rump length, as well reduced skeletal ossification (Hackett et al., 1982; Hardin et al., 1983). In a study with Fischer 344 rats, reduced fetal body weights were observed when the dams were exposed only during the period of organogenesis to 183 mg ethylene oxide/m3, a concentration having no overt toxic effects on the dams (Snellings et al., 1982a). Repeated brief exposures of pregnant Sprague-Dawley rats during gestation to 1464 or 2196 mg ethylene oxide/m3 produced a decline in fetal body weight (at both concentrations) and maternal toxicity (reduced body weight gain) at 2196 mg/m3 (Saillenfait et al., 1996); however, there was no evidence of teratogenicity.
In offspring of female hybrid mice exposed to 2196 mg ethylene oxide/m3 at various intervals shortly after mating, there was a range of congenital malformations, including omphalocoele, hydrops, eye defects, open thorax, cardiac defects, cleft palate and tail and limb defects (Generoso et al., 1987; Rutledge and Generoso, 1989). Increases in the numbers of mid-gestational and late fetal deaths and offspring that did not reach weaning were also observed (Generoso et al., 1987; Rutledge and Generoso, 1989; Rutledge et al., 1992). In the offspring of female mice exposed to ≥1647 mg ethylene oxide/m3 for brief periods shortly after mating, skeletal ossification was reduced and the incidence of axial skeletal anomalies and cleft sternum was increased (Polifka et al., 1991, 1992).
The intravenous administration of 150 and 18 mg ethylene oxide/kg-bw per day during gestation to mice and rabbits, respectively, produced fetotoxic effects as well as effects in the dams (LaBorde and Kimmel, 1980; Jones-Price et al., 1983). The intraperitoneal administration of a single dose of 125 mg ethylene oxide/kg-bw to pregnant hybrid mice during either the zygotic or the embryonic period reduced post-implantation survival and increased the incidence of skeletal abnormalities (information concerning maternal toxicity was not provided) (Polifka et al., 1996).
2.4.3.6 Neurological effects
Effects on the nervous system have been observed frequently in laboratory animals exposed to ethylene oxide. The paralysis observed in some animals was reversed upon the cessation of exposure (Hollingsworth et al., 1956). Poor coordination of the hind quarters was observed in rats and mice following exposure to 810 mg ethylene oxide/m3 for 7-8 weeks (Snellings, 1982). In subchronic or chronic studies in rats exposed to 458-915 mg ethylene oxide/m3, there was a range of neurological effects, including awkward or ataxic gait, 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 (Hollingsworth et al., 1956; Ohnishi et al., 1985, 1986; Matsuoka et al., 1990; Mori et al., 1990). Abnormal posture during gait and reduced locomotor activity were also observed in mice after exposure to ethylene oxide at concentrations ranging from 86 to 425 mg/m3, for 6 hours per day, 5 days per week, for 10 or 11 weeks (Snellings et al., 1984a); effects on various reflexes (righting, tail pinch, toe pinch) were also noted at the highest concentration examined (i.e., 425 mg/m3).
In a limited, poorly reported study in rabbits and monkeys, paralysis of the hind limbs was observed in both species accompanied by atrophy of the leg muscles, following exposure to ≥370 mg ethylene oxide/m3 for periods ranging from 7 to 32 weeks (exact exposure periods were not clearly specified) (Hollingsworth et al., 1956).
In two studies of male cynomolgus monkeys exposed to 92 or 183 mg ethylene oxide/m3 for 2 years (Sprinz et al., 1982; Lynch et al., 1984b), 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.
2.4.3.7 Toxicokinetics and mode of action
Information on the kinetics and metabolism of ethylene oxide has been derived primarily from studies conducted with laboratory animals exposed via inhalation, although some limited data from humans have been identified. In animals and humans, there are two routes of ethylene oxide catabolism, both of which are considered to be detoxification pathways. The first involves hydrolysis to ethylene glycol, with subsequent conversion to oxalic acid, formic acid and carbon dioxide. The second involves conjugation with glutathione, with subsequent metabolic steps yielding S-(2-hydroxyethyl)cysteine [S-(2-carboxymethyl)cysteine] and their N-acetylated derivatives (i.e., N-acetyl-S-(2-hydroxyethyl)cysteine [and N-acetyl-S-(2-carboxymethyl)cysteine]) (Wolfs et al., 1983; WHO, 1985; ATSDR, 1990; Popp et al., 1994). Based upon available data, the route involving conjugation with glutathione appears to predominate in rats and mice; in larger species (rabbits, dogs), the conversion of ethylene oxide is primarily via hydrolysis through ethylene glycol (Jones and Wells, 1981; Martis et al., 1982; Gérin and Tardif, 1986; Tardif et al., 1987).
Ethylene oxide is a substrate for the human GSTT1 enzyme (Hallier et al., 1993; Pemble et al., 1994; Hayes and Pulford, 1995). There appears to be no qualitative difference in the metabolism of ethylene oxide between laboratory animals (i.e., rodents) and humans. Brown et al. (1996) indicated that approximately 50% and 30% of total urinary metabolites arising from inhaled ethylene oxide would be derived via the glutathione conjugation pathway in mice and rats, respectively. Levels of GSTT1 enzyme activity decrease in the order mice > rats > humans.2 Brown et al. (1996) reported that the specific activity of GSTT1 (using ethylene oxide as the substrate) in the liver and kidney was greater in mice than in rats. The specific activity of GSTT1 (measured using dichloromethane as the substrate) in human tissues appears to be about 10% of that in mice (Reitz et al., 1989; Hashmi et al., 1994). In situ hybridization studies have indicated that differences in enzymatic activity are likely attributable (at least in part) to the differential expression of the GSTT1 gene among mice, rats and humans (Mainwaring et al., 1996).
Ethylene oxide is an electrophilic agent that alkylates nucleophilic groups in biological macromolecules, including DNA and proteins. In hemoglobin, for example, adducts can be formed at cysteine residues, N-terminal valine, as well as Nτ- and Nπ-histidine (Segerbäck, 1990). Since ethylene oxide is formed during the metabolism of ethylene, a natural body constituent, endogenous as well as exogenous sources of ethylene and ethylene oxide contribute to background alkylation of proteins such as hemoglobin and albumin, as well as DNA (Bolt, 1996). N-(2-hydroxyethyl)valine (HEVal) and hydroxyethylhistidine (HEHis) adducts have been frequently monitored in tissues of workers exposed to ethylene oxide in occupational settings (see IARC, 1994). Background levels of HEVal in non-smokers ranged from 9 to 188 pmol/g globin (Törnqvist et al., 1986, 1989; Bailey et al., 1988; Hagmar et al., 1991; Sarto et al., 1991; Tates et al., 1991, 1992; van Sittert et al., 1993; van Sittert and van Vliet, 1994; Farmer et al., 1996; Granath et al., 1996). Ethylene oxide binding to DNA results primarily in the formation of 7-(2-hydroxyethyl)guanine (7-HEGua) (Föst et al., 1989; Li et al., 1992); other adducts have also been identified at much lower levels. In DNA extracted from the lymphocytes of unexposed individuals, mean background levels of 7-HEGua ranged from 2 to 8.5 pmol/mg DNA (Föst et al., 1989; Bolt et al., 1997). Although these levels were similar to those measured in rodents not exposed to ethylene oxide (Föst et al., 1989; Walker et al., 1992), Wu et al. (1999a), using a more sensitive technique, recently reported that human tissue contains 10- to 15-fold higher levels of endogenous 7-HEGua than rodent tissue.
Among rodent species, there are clear quantitative differences in metabolism. Clearance of ethylene oxide from the blood (and other tissues) was approximately 3- to 4-fold faster in mice than in rats; peak levels of ethylene oxide in the muscle and brain of rats and mice were similar (Brown et al., 1996). GSTT1 activity was highest in the liver, followed by the kidney and testes. Rat brain and rat and mouse lung contained small amounts of activity, compared with other tissues (enzyme activity in mouse brain was not examined).3 Levels of ethylene oxide in the testes of rats were 20% of those in other tissues, while in mice, levels in the testes were 50% of those measured in other tissues (Brown et al., 1996). In mice, half-lives for the elimination of 7-HEGua in DNA from a variety of tissues (brain, lung, spleen, liver and testes) were 1.5- to 3.9-fold lower than in rats (Walker et al., 1992). In both rats and mice, substantive depletion of glutathione pools has been observed following acute exposure to high levels (i.e., ≥300 ppm [549 mg/m3]) of ethylene oxide (McKelvey and Zemaitis, 1986; Brown et al., 1998), although it should be noted that increases in tumour incidence have been observed at lower concentrations.
It is likely that toxicological effects associated with exposure to ethylene oxide arise primarily as a result of its direct alkylation of biological macromolecules (i.e., nucleic acids, proteins). in vivo exposure to ethylene oxide induced mutations (5- to 5.6-fold) at the Hprt locus in splenic T-lymphocytes in rats and mice (Walker et al., 1997a,b). A statistically significant (i.e., p < 0.05) increase (1.5-fold) in the frequency of lacI mutation was observed in the lungs of transgenic mice exposed to 200 ppm (366 mg/m3) ethylene oxide (Sisk et al., 1977); the increased frequency of lacI mutations in the bone marrow and spleen of these animals (1.9- and 1.3-fold, respectively) was not statistically different from the unexposed controls. Currently, there is no clear evidence of a relationship between the mutagenic response observed at these two "indicator" loci and the species- and tissue-specific carcinogenicity of ethylene oxide. Molecular analysis of ethylene oxide-induced mutations at the HPRT locus in human diploid fibroblasts exposed in vitro revealed that a high proportion involved large deletions of this gene (Bastlová et al., 1993).
A potential role of the formation of 7-HEGua in contributing to the carcinogenic response has been the focus of many studies, this adduct having been identified in both humans and laboratory animals. In reports by Walker et al. (1992) and Wu et al. (1999b), F344 rats and B6C3F1 mice were exposed (via inhalation for 6 hours per day, 5 days per week, for 4 weeks) to concentrations of ethylene oxide similar to those used in previous carcinogenicity bioassays involving these strains (Lynch et al., 1984a,b; Snellings et al., 1984b; Garman et al., 1985; Garman and Snellings, 1986; NTP, 1987). Slightly higher levels of 7-HEGua were measured in tissues (lung, spleen, brain, liver) from rats than from mice; within each species, similar levels of the adduct were measured in the lung, spleen, brain and liver. Since an increased incidence of brain tumours has been observed in rats but not in mice exposed to ethylene oxide and an increased incidence of lung tumours has been observed in mice but not in rats exposed to this substance, the results provided by Walker et al. (1992) and Wu et al. (1999b) point to no obvious relationship between the overall level of 7-HEGua within various tissues and the observed species-specific carcinogenic response. The potential roles of this and/or other ethylene oxide-induced DNA adducts as well as other factors in mediating the carcinogenicity of ethylene oxide have not been defined.
2.4.4 Humans
2.4.4.1 Non-neoplastic effects
2.4.4.1.1 Irritation and sensitization
Exposure to ethylene oxide vapour can cause irritation of the eyes, nose and throat; however, the sensory warning signs of the chemical are poor (ATSDR, 1990). Brief exposure of workers to the chemical produced irritation of the mucous membranes of respiratory passages, which led to bronchitis, pulmonary edema and, in one case, emphysema (Thiess, 1963). Mild irritation of the skin has been reported after contact with aqueous solutions of ethylene oxide as low as 1% (Sexton and Henson, 1949). Dermal injury is characterized by edema and erythema, occurring 1-5 hours after exposure, followed by the formation of vesicles. The magnitude of the injury is dependent upon both the duration of contact and the concentration to which the individual is exposed (WHO, 1985). Dermal irritation has also been observed after contact with ethylene oxide-sterilized materials and clothing, such as surgical dressings and tubing, face masks, gloves, boots, surgical gowns and overalls (Royce and Moore, 1955; Marx et al., 1969; Hanifin, 1971; Biro et al., 1974; LaDage, 1979; Bommer and Ritz, 1987; Fisher, 1988; Lerman et al., 1995).
Ethylene oxide is considered a good sensitizing agent, owing to its strong reactivity with various chemical groups. 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 (e.g., hemodialysis, peritoneal dialysis, plasmapheresis, plateletpheresis) involving equipment sterilized by exposure to ethylene oxide (reviewed in Bommer and Ritz, 1987). Asthmatic reactions may occur either alone or in combination with anaphylactic events; case reports of occupational asthma attributed to ethylene oxide exposure have appeared (Dugue et al., 1991; Verraes and Michel, 1995). Reports of contact dermatitis attributed to ethylene oxide are not uncommon; however, the clinical diagnosis should distinguish a true allergic reaction from the normally irritative effects of this chemical.
2.4.4.1.2 Reproductive effects
In studies of female sterilization workers and one study involving paternal exposure to ethylene oxide, increased risks of adverse pregnancy outcome, most often spontaneous abortion, have been reported. In one study, Hemminki et al. (1982) determined the incidence of spontaneous abortion among Finnish hospital staff who had used ethylene oxide, glutaraldehyde and formaldehyde for instrument sterilization. All sterilizing staff employed in Finnish hospitals in 1980 were included in the analysis, with a total of 1443 pregnancies (545 workers exposed during pregnancy). Measurements carried out in 24 Finnish hospitals since 1976 revealed 8-hour, time-weighted-average (TWA) exposures ranging from 0.1 to 0.5 ppm (0.2 to 0.9 mg/m3) ethylene oxide, with peak concentrations up to 250 ppm (458 mg/m3). However, no measurements were taken during the study, and the concentrations of ethylene oxide may have been higher prior to 1976. When the pregnancies of the sterilizing staff were analysed according to employment at the time of conception, the rate of spontaneous abortion was significantly (p < 0.001) increased in the exposed (15.1%) versus the unexposed (4.6%) group. When the associations between ethylene oxide and the different sterilizing agents were analysed, only exposure to ethylene oxide during early pregnancy was related to an increased frequency of spontaneous abortion (adjusted rate of 16.1% in exposed versus 7.8% in unexposed workers; p < 0.01). Hospital discharge records revealed a similar pattern, with spontaneous abortion rates of 22.6% (significantly higher than controls, p < 0.05), 9.9% and 9.2% in sterilizing workers exposed to ethylene oxide, unexposed workers and controls, respectively. In a subsequent analysis, only pregnancies that began during hospital employment were analysed in all groups, with controls chosen from the same hospitals (Hemminki et al., 1983). The rate of spontaneous abortion remained significantly higher (p < 0.05) among the pregnancies associated with exposure to ethylene oxide (20.4%) compared with controls (11.3%).
Rowland et al. (1996) examined the occurrence of spontaneous abortion and pre- and post-term delivery in relation to ethylene oxide exposure among 7000 randomly selected dental assistants (aged 18-39) identified from the 1987 dental assistant registry in California. The most recent pregnancy outcome was chosen for analysis to maximize recall of pregnancy and exposure information, with 1320 women who provided information on age and ethylene oxide exposure contributing to the analysis. A total of 32 women reported ethylene oxide exposure during pregnancy; no quantitative measures or details on timing of exposure during pregnancy were available. The age-adjusted relative risk of spontaneous abortion among ethylene oxide-exposed women was 2.5 (95% confidence interval [CI] = 1.0-6.3); the relative risks of pre-term births (21-37 weeks) and post-term births (≥42 weeks) were 2.7 (95% CI = 0.8-8.8) and 2.1 (95% CI = 0.7-5.9), respectively. Using a logistic model, ethylene oxide-exposed women were 2.7 times (95% CI = 1.2-6.1) more likely to have any of the three adverse pregnancy outcomes after adjusting for age. Adjustment for unscavenged nitrous oxide exposure, high amalgam use and smoking yielded a relative risk of 2.1 (95% CI = 0.7-5.7). In the only identified study in which the effect of paternal exposure to ethylene oxide on reproductive outcome was assessed, Lindbolm et al. (1991) reported a significantly (p < 0.05) increased risk of spontaneous abortion (odds ratio [OR] = 4.7; 95% CI = 1.2-18.4) among Finnish women whose partners had been exposed to ethylene oxide. In total, 99 186 pregnancies were included in the analysis. Paternal exposure to ethylene oxide was based upon the job and industry in which the men were employed; quantitative data on exposure were not available, and the numbers of spontaneous abortions (n = 3) and pregnancies (n = 10) in the paternal ethylene oxide-exposed group were small. Other potential confounding factors, such as previous abortions and alcohol and tobacco consumption, were not considered in the analysis.
2.4.4.1.3 Neurological effects
Sensorimotor polyneuropathy was reported for a number of cases following acute or chronic exposure to ethylene oxide (exposure concentrations, when reported, ranged from 4.2 to >700 ppm [7.7 to >1281 mg/m3]) (Gross et al., 1979; 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). Amelioration of the symptoms following the cessation of exposure has been commonly observed. Abnormally low nerve conduction velocities were observed in some cases, along with numbness and weakness in the extremities, slow and clumsy alternating hand movements, decreased muscle stretch reflexes in extremities, heel-shin ataxia and unsteady and wide-based gait. 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). Effects on the central nervous system (e.g., seizures) have been observed following acute exposure to 500-700 ppm (915-1281 mg/m3) ethylene oxide (Gross et al., 1979; Salinas et al., 1981).
Chronic occupational exposure to ethylene oxide has been associated with impaired performance in neuropsychological testing (most often on tests of psychomotor skills) and, in one study, with reduced peripheral nerve conduction velocity. However, in all identified studies, the numbers of subjects, often of unknown comparability, were small. In eight females who had worked at a hospital with or in the proximity of ethylene oxide gas sterilizers for 5-20 years (mean 11.6 years; TWA in breathing zone up to 3 ppm [5.5 mg/m3]), performance was poor on all psychometric tests and significantly less accurate in the hand-eye coordination test (p = 0.03) than that in unexposed hospital workers or members of the same union matched for age and sex (Estrin et al., 1987). A significant correlation was observed between decreasing performance on the continuous performance test and years of exposure to ethylene oxide, after controlling for age (r = 0.67; ≤ 0.05) (Estrin et al., 1987). Subsequently, in 10 female hospital workers with chronic exposure to ethylene oxide (breathing zone levels up to 250 ppm [458 mg/m3]), the proportion of individuals with bilaterally reduced ankle reflexes was higher (≤ 0.05) than in the same number of unexposed controls (Estrin et al., 1990). There was no difference with respect to sural and peroneal amplitudes and velocities, nor were there any significant differences on neuropsychological tests of cognition requiring verbal ability. Compared with the controls, the ethylene oxide-exposed group performed poorly on the Trails A test (p = 0.04), the cancellation test (p = 0.06) and the computerized finger tapping test (p = 0.009), all of which assess psychomotor skills. The exposed group had a statistically significant lower P300 amplitude, which has been associated with cognitive dysfunction (Estrin et al., 1990).
Neuropsychological effects possibly associated with chronic exposure to ethylene oxide were examined in a cross-sectional study of 22 male and female hospital supply workers exposed to low levels of ethylene oxide and in 24 unexposed controls from local hospitals (Klees et al., 1990). The 8-hour TWA concentration was 4.7 ppm (8.6 mg/m3) ethylene oxide. Significantly more individuals in the ethylene oxide-exposed group (p < 0.05) were classified (by two neuropsychologists) as "impaired" (n = 5) compared with the controls (n = 1), following evaluation using a neuropsychological screening battery designed to assess a spectrum of neuropsychological functions (including memory, verbal, visual-spatial and psychomotor functions) considered likely to be affected by neurotoxic exposures.
2.4.4.1.4 Genetic effects
Increases in chromosomal aberrations in peripheral blood lymphocytes have been consistently reported in studies of workers exposed to concentrations of ethylene oxide of ≥5 ppm (9.2 mg/m3) (Table 4). Effects observed at lower concentrations (i.e., <5 ppm [9.2 mg/m3]) have been mixed.
Significant increases in the frequency of sister chromatid exchange have also been observed among individuals exposed to elevated levels of ethylene oxide (i.e., usually ≥5 ppm [9.2 mg/m3]). Studies of individuals exposed to lower levels (i.e., <0.5 ppm [0.9 mg/m3]) have yielded mixed results. In some studies, increases in the frequency of sister chromatid exchange have been observed to persist after exposure had ceased. Effects related to the concentration or duration of exposure to ethylene oxide have been observed in a number of studies.
In some studies, the frequency of micronuclei in peripheral blood was increased in workers exposed to relatively high (2-33 ppm [3.7-60.4 mg/m3]) levels of ethylene oxide (Tates et al., 1991; Ribeiro et al., 1994). However, in the majority of the studies involving exposures to lower levels, no effect on the frequency of micronuclei was observed. Apparent inconsistencies in the data could reflect the influence of peak exposures, differences in exposure duration or smoking status.
Table 4 Cytogenetic effects in humans (modified from IARC, 1994)
In a study involving small numbers (i.e., n = 4-12 per group, depending upon exposure group) of non-smoking males and females exposed to ethylene oxide through the sterilization of medical equipment, Fuchs et al. (1994) reported statistically significant (p < 0.05), concentration-dependent increases (1.5- and 2.2-fold, respectively) in single-strand DNA breaks in peripheral mononuclear blood cells obtained from individuals exposed to (4-hour TWA) ethylene oxide concentrations of 0.1-0.49 mg/m3 and 0.5-2.0 mg/m3, compared with workers exposed to levels below 0.1 mg/m3. A 1.5-fold increase in DNA single-strand breaks in cells from workers exposed to >2 mg ethylene oxide/m3 was not significantly different (i.e., p > 0.05) from the number observed in cells from workers exposed to <0.1 mg/m3. The group of non-smokers could be divided into subgroups having a "higher" or "lower" sensitivity to ethylene oxide.
2.4.4.1.5 Other non-neoplastic effects
Only a limited range of other non-neoplastic effects has been investigated in humans exposed to ethylene oxide. Hematological changes were observed among a group of 59 women exposed to ethylene oxide released from sterilizers while employed at nine hospitals in the United States and one in Mexico (Schulte et al., 1995). Exposure was classified as none, low or high, based on mean 4-month cumulative exposure categories of 0, >0-32 or >32 ppm-hour, respectively. Mean 8-hour TWA exposures in the U.S. hospitals were 0.08 ppm (0.15 mg/m3) (range = 0-0.55 mg/m3) and 0.17 ppm (0.31 mg/m3) (range = 0.24-0.55 mg/m3) for the low- and high-exposure categories, respectively; the corresponding measurements in the Mexican hospital were 0.02 ppm (0.04 mg/m3) and 0.54 ppm (0.99 mg/m3) (range = 0.5-2.5 mg/m3), respectively. Among the U.S. workers, hematocrit and hemoglobin levels were reduced (not statistically significant) in the high-exposure group, compared with the unexposed controls; the levels were significantly lower in the high-exposure group when compared with the low-exposure group (p = 0.03 and 0.02 for hemoglobin and hematocrit levels, respectively). Compared with unexposed controls, U.S. workers in the high-exposure group exhibited a statistically significant (p = 0.04) increase in the percentage of lymphocytes and a reduction (p = 0.03) in the percentage of neutrophils in the blood. Among the Mexican workers, an exposure-related increase (not statistically significant) in the percentage of neutrophils in the blood was observed.
Hematological changes were not observed either in a group of 36 male workers employed at an ethylene oxide manufacturing plant with estimated 8-hour TWA exposures below 0.05 ppm (0.09 mg/m3) (van Sittert et al., 1985) or in a group of 84 male workers involved in the manufacture of ethylene oxide who were exposed to estimated concentrations of <1 ppm (1.83 mg/m3) (Currier et al., 1984).
The prevalence of lens opacities and cataracts was assessed in a group of 55 workers exposed to ethylene oxide at five hospitals in Paris, France (Deschamps et al., 1990). Airborne concentrations of ethylene oxide ranged from 0.06 ppm (0.11 mg/m3) during a 97-minute exposure to 39 ppm (71 mg/m3) during a 2.5-minute exposure. Initial results revealed a greater prevalence of lens opacities among individuals over 45 years of age, and therefore comparisons were limited to this age group. There was no difference between the exposed and control groups in the prevalence (19/21 exposed; 10/16 unexposed), location, importance or type of lens opacities observed. Cataracts were observed in six exposed individuals (all over 45 years of age except one who was 44 years old), compared with none in the control group (p < 0.05).
2.4.4.2 Cancer
Associations between occupational exposure to ethylene oxide and various types of cancer have been examined in a number of epidemiological studies. A summary of the risk measures for selected cancers (stomach, pancreas, brain, hematopoietic system) is presented in Table 5.
In a cohort study of 709 ethylene oxide production and sterilization workers in Sweden, mortality due to leukemia (standardized mortality ratio [SMR] = 921; 7 observed deaths), blood and lymphatic cancer (SMR = 459; 9 observed deaths) and stomach cancer (SMR = 546; 10 observed deaths) was increased (Hogstedt, 1988). The greatest excess mortality was observed in operators and repairmen employed in an old production plant where ethylene oxide had been synthesized by the chlorohydrin method from 1941 to 1947 (mostly in an enclosed building), although there were no trends by duration of employment (<10 years, >10 years). Ethylene oxide exposure levels in all of the facilities were estimated to have been relatively high in the early years (average exposure levels were estimated to have been 14 ppm [26 mg/m3] between 1941 and 1947 at the plant producing ethylene oxide by the chlorohydrin method; however, peaks above the odour threshold of 400 ppm [732 mg/m3] had been reported).
Greenberg et al. (1990) conducted a study of 2174 workers employed at two ethylene oxide production plants in the United States, for which a 10-year update of this cohort, which excluded 278 chlorohydrin workers, was reported by Teta et al. (1993). Comparisons were with both the general population and unexposed workers in the plants. In this cohort, there were no statistically significant increases in deaths from any cause for the entire cohort; SMRs for cancer of the stomach, pancreas, brain and nervous system, and leukemia and aleukemia were 160 (95% CI = 69-315; 8 observed deaths), 61 (95% CI = 17-156; 4 observed deaths), 150 (95% CI = 55-327; 6 observed deaths), and 106 (95% CI = 35-248; 5 observed deaths), respectively (Teta et al., 1993). Although no increased mortality was observed among men from the high-exposure departments (expected number of deaths not reported), a statistically significant excess of deaths due to stomach cancer was observed in the intermediate-exposure group (SMR = 364; 95% CI = 102-957; 4 observed deaths); an increase (not statistically significant) was also observed in the low-exposure group (SMR = 222; 95% CI = 61-575; 4 observed deaths). When risks associated with duration of assignment were examined, there were no significant trends for all cancers, leukemia, pancreatic, brain or stomach cancers, although the numbers of deaths from cancer at any site were small. However, the relative risk for stomach cancer (2.77; 95% CI = 1.11-6.93; 5 observed deaths) was significantly elevated for those exposed from 2 to 9 years. An analysis of the 278 ethylene chlorohydrin production workers, considered to have only low, intermittent exposure to ethylene oxide, revealed SMRs for deaths due to digestive and peritoneal, pancreatic, respiratory and lymphatic/hematopoietic cancers of 143 (95% CI = 74-250; 12 observed deaths), 492 (95% CI = 158-1140; 8 observed deaths), 136 (95% CI = 76-224; 15 observed deaths) and 294 (95% CI = 127-580; 8 observed deaths), respectively (Benson and Teta, 1993).
TWA exposures were estimated over four time periods and three exposure intensity categories (low, medium, high). Average exposures in the most recent time period were based on industrial hygiene monitoring conducted in the plants and were inferred for earlier time periods based on exposure levels in similar manufacturing operations during the period of interest. A separate age-dependent exposure history was developed for each worker, based on the worker's assignments and estimated exposure levels (i.e., low, medium, high).
While there were no quantitative estimates of individual exposure in this investigation (assignments were to low, medium or high exposure only), the length of follow-up was among the longest in any study (mean duration of follow-up, 27.2 years; mean length of exposure, 5.4 years). Workers were exposed to a variety of other chemicals (approximately 26, including butadiene and benzene) (Shore et al., 1993).
Table 5 Summary of risk measures for selected cancers (stomach, pancreas, brain, hematopoietic) from epidemiological studies
In the largest single cohort studied to date, Steenland et al. (1991) examined mortality in 18 254 male and female workers exposed to ethylene oxide at 14 plants producing sterilized medical supplies and spices in the United States. Comparisons were with the U.S. general population. A more detailed analysis of exposures in this cohort was subsequently conducted by Stayner et al. (1993), being restricted to workers from 13 of the 14 original facilities having adequate information for estimating historical exposures. The database on which these exposure estimates were based was collected in the course of walkthrough surveys of 36 companies in the medical supplies and spice industries and in-depth sampling surveys of 2 of these 36 companies; the database included 2350 individual TWA exposure values acquired from 18 facilities between 1976 and 1985 (Greife et al., 1988). Arithmetic mean exposures were calculated by facility, year and exposure categories. The exposure categories were based on grouping of all sampled jobs into eight categories with similar potential for exposure. An industrial hygiene-based regression model was used to estimate exposure to ethylene oxide for each exposure category. This model predicted exposures to e thylene oxide within 1.1 ppm (2.0 mg/m3) of a validation data set (46 measurements not used in the model), with a standard deviation of 3.7 ppm (6.8 mg/m3) (Hornung et al., 1994). Cumulative exposure for each individual was estimated by integration of estimated ppm ethylene oxide for each job held multiplied by the duration of time (days) spent in the job.
For the entire cohort, there was no increase in mortality from hematopoietic cancer. There was a slight but significant increase among men, however, but a decrease among women (Steenland et al., 1991). The SMR for deaths due to "all haematopoietic neoplasms" among the group with the highest cumulative exposure was 124 (not statistically significant; 95% CI = 66-213), and the trend with cumulative exposure was not statistically significant. For the group with the highest cumulative exposure, the SMRs for non-Hodgkin's lymphoma and leukemia/aleukemia were 192 (95% CI = 77-395) and 75 (95% CI = 15-218), respectively. Increased mortality from kidney cancer (SMR = 322; 95% CI = 139-635) was observed in the mid-cumulative exposure group; however, no trend with exposure was observed (Stayner et al., 1993).
When the results for "all haematopoietic neoplasms" were stratified according to sex, increased mortality (SMR = 196; 95% CI = 101-343) was observed among males in the highest exposure category, and there was a suggestive positive trend in the SMRs with exposure. When results for all workers were stratified according to time since first exposure, the greatest excess in deaths due to "haematopoietic neoplasms" was observed among workers with more than 20 years since their first exposure (SMR = 155; 95% CI = 77-277). Regression analysis revealed a highly statistically significant exposure-response relationship between cumulative exposure to ethylene oxide and mortality from lymphocytic leukemia and non-Hodgkin's lymphoma combined (termed "lymphoid" neoplasms and combined based on the consideration that these neoplasms may represent different expressions of the same underlying disease process). A marginal relationship was also observed between cumulative exposure and mortality from all hematopoietic neoplasms and from non-Hodgkin's lymphoma. There was a positive, but not statistically significant, exposure-response relationship between cumulative exposure and leukemia. Inclusion of a 5-year lag period yielded a stronger exposure-response relationship for "lymphoid" neoplasms; the relationship between cumulative exposure and mortality from all hematopoietic neoplasms and from non-Hodgkin's lymphoma was statistically significant, after inclusion of a 10-year lag period in the analysis. A negative exposure-response relationship was observed between cumulative exposure and cancers of the stomach, pancreas, brain and kidney (Stayner et al., 1993).
While this is one of the few studies for which there are individual estimates of cumulative exposure, the monitoring data on which these estimates are based are limited to those collected after 1978 (however, results for the subgroup of this cohort exposed before 1978 were essentially identical to those for the entire cohort, suggesting that exposures may have been similar). Exposure to ethylene oxide was, however, relatively unconfounded by concomitant exposures to other substances, with none being identified. Although this is by far the largest of all of the studies conducted to date, the average follow-up period was short; 28% of workers had attained >20 years since first exposure (average duration of follow-up, 16 years; mean duration of exposure, 4.9 years). In this regard, it should be noted that five of the seven men who died of leukemia died in the most recent calendar period, generating statistically significant excess mortality for these years (SMR = 345; 95% CI = 111-806), and additional follow-up of this cohort is desirable. The variation in response between men and women in this study cannot be explained on the basis of sex ratio of the study population (i.e., small numbers of women); indeed, a greater proportion of the population was female (55% versus 45%).
In a subsequent analysis of data from the cohort evaluated by Steenland et al. (1991), in which the categorization of non-Hodgkin's lymphoma included an additional International Classification of Disease category that was omitted in the original Steenland et al. analysis (i.e., "other neoplasms of lymphoid tissue"), Wong and Trent (1993) reported increased mortality due to non-Hodgkin's lymphoma in males (SMR = 247; 95% CI = 141-402). No indication of an exposure-response relationship associated with either duration of employment or latency was observed. Similarly, there was no mortality pattern by latency or duration of employment for any cancer site examined. While this study is slightly larger than the corresponding study by Steenland et al. (1991), individual estimates of exposure frequency and intensity were not assigned.
Cancer risks were not significantly increased in a cohort of 2170 male and female Swedish workers exposed to ethylene oxide at two plants producing disposable medical equipment (Hagmar et al., 1995). The risk of lymphopoietic/hematopoietic cancers was elevated (standardized incidence ratio [SIR] = 1.78; 95% CI = 0.65-3.88; 6 observed cases); two of the cases had leukemia (SIR = 2.44; 95% CI = 0.3-8.81). When the analysis excluded workers with cumulative exposures to ethylene oxide below the median value of 0.13 ppm-years, but included a minimum 10-year latency period, there was an increased (although not statistically significant) risk of leukemia (SIR = 7.14; 95% CI = 0.87-25.8; 2 observed cases). Cases with leukemia had a slightly higher cumulative exposure to ethylene oxide than the average cohort member. In this study, the levels of adducts in hemoglobin correlated well with the estimated exposure levels (Hagmar et al., 1995). This is one of the only studies in which there were cumulative estimates of individual exposure; moreover, biological dosimetry of hemoglobin adducts was conducted to corroborate exposure estimates. Exposures were also mainly to ethylene oxide; the only other exposures were to methyl formate or fluorochlorocarbons. However, the mean duration of follow-up was relatively short (11.6 years), and exposure levels were relatively low for most workers, since fewer than 200 workers had more than 1 ppm-year of cumulative exposure.
A nested case-control study of 10 cases of Hodgkin's disease among male employees of a Belgian chemical manufacturing firm revealed a statistically significant increased risk associated with exposure to ethylene oxide (OR = 8.5; 95% CI = 1.4-39.9; 3 exposed cases) (Swaen et al., 1996). The risk remained significantly elevated after restricting the analysis to individuals with durations of exposure of more than 10 years. The expected number of cases was 2.01 (SIR = 497; 95% CI = 238-915).
No statistically significant increased risks of cancer of the hematopoietic system or other sites were observed in a number of other epidemiological studies, in which there were no estimates of individual exposure and small numbers of hematopoietic and other cancers (Morgan et al., 1981; Gardner et al., 1989; Kiesselbach et al., 1990; Olsen et al., 1997). In the study by Kiesselbach et al. (1990), there were no increases in leukemias or total hematopoietic cancers, nor were there any trends associated with exposure intensity, duration of exposure or latency in 3658 men from six chemical companies, with median length of exposure of 9.6 years. Gardner et al. (1989) reported a small excess of leukemia mortality among chemical workers (3 observed/1.33 expected) and a deficit among hospital workers (0 observed/0.76 expected), although neither was statistically significant, in 2876 men and women from four companies that produced or used ethylene oxide and eight hospitals that used ethylene oxide sterilizers (average duration of follow-up not reported). Morgan et al. (1981) observed non-significant excesses of brain cancer, pancreatic cancer and Hodgkin's disease but a deficit of leukemia in 767 men at an ethylene oxide production plant. For the entire cohort examined by Olsen et al. (1997), there were non-statistically significant excess deaths from cancer of the large intestine, lung, kidney, lymphopoietic/ hematopoietic tissue and other lymphatic tissues. Among those involved only in the production of ethylene chlorohydrin, the SMR was increased for lymphopoietic/hematopoietic cancer (SMR = 149; 95% CI = 60-307; 7 observed/4.7 expected).
In two additional small studies, significantly increased risks of lymphosarcoma/ reticulosarcoma and breast cancer have been observed by Bisanti et al. (1993) and Norman et al. (1995), respectively. In the former investigation, the SMR for lymphosarcoma/ reticulosarcoma was 682 (4 observed deaths, p < 0.05) among 1971 chemical workers licensed to handle ethylene oxide between 1938 and 1984 in Italy, in comparison with the local population (mean length of follow-up, 9.8 years; no estimates of individual exposure). However, there was no association with duration or latency, although the former could not be accurately determined.
A meta-analysis of results from 13 epidemiological studies published between 1979 and 1993 was conducted by Shore et al. (1993). The magnitude and the consistency (heterogeneity) of the relative risks were evaluated for the individual and combined studies, along with any trends associated with intensity or frequency of exposure, duration of exposure or time since first exposure (latency) for the cancers of greatest interest - namely, cancers of the pancreas, brain, stomach, leukemia and non-Hodgkin's lymphoma. For leukemia, SMRs from the individual studies were heterogeneous, due exclusively to the results reported by Hogstedt (1988). The summary SMR (sSMR) was 1.06 (95% CI = 0.73-1.48); when accounting for heterogeneity, the SMR was 1.06 (95% CI = 0.55-2.02) (Shore et al., 1993). There was no consistency in the trends associated with the frequency and intensity of exposure for the individual studies, nor were any of the trends statistically significant. No trend with respect to frequency, intensity or duration of exposure was observed for the risk of leukemia for all studies combined; however, there was a suggestion that risks increased with the latency period. The sSMR for non-Hodgkin's lymphoma was not significantly increased (1.35; 95% CI = 0.93-1.90); the test for heterogeneity was not statistically significant. 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 study was noted. No trends with respect to duration of exposure or latency were noted. There was a lack of homogeneity among the relative risks for stomach cancer, due to the results of one study reported by Hogstedt (1988), in which the SMR was 7.1. The overall sSMR for stomach cancer for all of the studies, taking this heterogeneity into account, was 1.28 (95% CI = 0.73-2.26). No trends associated with the intensity or duration of exposure, latency or cumulative exposure were observed. The sSMRs for pancreatic cancer (0.98; 95% CI = 0.69-1.36), brain and nervous system cancer (0.89; 95% CI = 0.39-2.04, taking into account heterogeneity) and all cancers (0.94; 95% CI = 0.88-1.01) were not increased, nor were any trends observed (Shore et al., 1993). No attempt was made to weight the studies according to quality, although there was a narrative critique of each. The authors also noted that data on exposure to ethylene oxide were inadequate in most of the studies, but that the cumulative exposure analysis by Stayner et al. (1993) represented a significant advance in the quantitative analysis of effects of ethylene oxide.
Teta et al. (1999) reported on an update of the Shore et al. (1993) meta-analysis, using methods and studies similar to those employed by Shore et al. (1993), but also including data from Hagmar et al. (1995) and Olsen et al. (1997). Similar to the findings reported by Shore et al. (1993), the meta-SMRs (mSMRs) for non-Hodgkin's lymphoma, leukemia, and cancer of the pancreas, brain and stomach were 1.34 (95% CI = 0.96-1.89), 1.08 (95% CI = 0.61-1.93), 0.95 (95% CI = 0.69-1.31), 0.96 (95% CI = 0.49-1.91) and 1.23 (95% CI = 0.71-2.13), respectively. Reportedly, there were no statistically significant trends with respect to duration or intensity of exposure (data not shown). There was a reportedly statistically significant (p < 0.05) trend with respect to latency and brain cancer, based on four studies (data not shown).
1Mononuclear cell leukemias are unique to the F344 strain of rat. These tumours arise spontaneously, primarily in older animals. The exact etiology of this tumour type, including cell of origin, has not been definitively identified.
2Although sites at which tumours occur in rats and mice vary, carcinogenic potency is generally greater in rats than in mice.
3These results are consistent with the observation of tumours in mouse lung and rat brain, if GSTT1 activity is the critical determinant, but not with the lack of observation of tumours in rat lung.
