Data on sources and emissions from Canada, the source country of the national assessment on which this CICAD is based, are presented here as an example. Sources and patterns of emissions in other countries are expected to be similar, although quantitative values may vary.

Estimates of emissions of butadiene are highly variable, depending on the method of estimation and the quality of the data upon which they are based. Total Canadian emissions for 1994 were estimated to range between 12 917 and 41 622 tonnes (Environment Canada, 1998). Major uncertainties are associated with estimates for combustion sources, notably forest fires.

Butadiene is released from biomass combustion, especially forest fires. Total global emissions of buta diene from biomass combustion were estimated to be 770 000 tonnes per year (Ward & Hao, 1992). Releases from forest fires in Canada were estimated to range between 3607 and 26 966 tonnes, which constituted 49.3% (range of estimates is 28–65%) of the total annual emissions of butadiene in Canada (CPPI, 1997). Although Altshuller et al. (1971) suggested that buta diene can be released from natural gas losses and diffu sion through soil from petroleum deposits, no data were identified on this possible source.

All internal combustion engines may produce butadiene as a result of incomplete combustion. The amount generated and released depends primarily on the composition of fuel, the type of engine, the emission control used (i.e., presence and efficiency of catalytic converter), the operating temperature, and the age and state of repair of the vehicle.² Cyclohexane, 1-hexene, 1-pentene, and cyclohexene have been identified as primary fuel precursors for butadiene (Schuetzle et al., 1994). As well, very low levels of butadiene itself may be present in gasoline and in liquefied petroleum gas.

Butadiene can also enter the environment from any stage in the production, storage, use, transport, or dispo sal of products with residual, free, or unreacted butadi ene. Data on Canadian industrial emissions have been collected for industrial processes, plastic products indus tries, refined petroleum and coal products industries, and chemical and chemical products industries as part of the National Pollutant Release Inventory (NPRI) (Environ ment Canada, 1996a, 1997). Emissions other than those reported to the NPRI may occur, including from combus tion of other fuels (e.g., natural gas, oil, and wood space heating), prescribed forest burning, cigarettes, waste incineration, releases from polymer products, releases from the use and disposal of products containing buta diene, and spillage (Ligocki et al., 1994; Environment Canada, 1996b; OECD, 1996).

The following amounts of butadiene were estimated to have been released into the Canadian environment in 1994 from key transportation and related sources (Environment Canada, 1996a; CPPI, 1997): 3376–7401 tonnes from on-road gasoline- and diesel- powered motor vehicles (with about 45–89% of those releases from gasoline engines and 11–55% from diesel engines); 150–258 tonnes from aircraft; 84–1689 tonnes from off-road motor vehicles; 84 tonnes from lawn mowers; 40 tonnes from the marine sector; and 17 tonnes from the rail sector.

In addition, data from NPRI for 1994 (Environ ment Canada, 1996a) listed a total of 270.4 tonnes released from the chemical and chemical products industries. Of this, 270.3 tonnes were released into air, 0.058 tonnes into water (St. Clair River, Ontario), and 0.002 tonnes onto land. There were releases of 17.5 tonnes into air from the plastic products industries. A total of 22.3 tonnes was released from the refined petroleum and coal products industries, of which 22.2 tonnes were released into air. Off-site transfer of wastes (material sent for final disposal or treatment prior to final disposal) from industrial sites in Canada in 1994 was estimated to include a total of 131.3 tonnes of buta diene, with 128.7 tonnes being sent to incineration, 2.1 tonnes to landfill, and 0.5 tonnes to municipal sewage treatment plants (Environment Canada, 1996a). Based on 1995 NPRI data (Environment Canada, 1997), the amount of butadiene estimated to have been released into the Canadian environment was 225.8 tonnes from industrial on-site uses, with 0.058 tonnes released into water, 0.002 tonnes into land, and 225.4 tonnes into air. Releases into air included air fugitive releases (172.8 tonnes), air stack releases (36.3 tonnes), air stor age releases (4.8 tonnes), air spill releases (1.1 tonnes), and other air releases (10.4 tonnes).

Based on data in NPRI, it was estimated that the total release of butadiene from fuel distribution in 1994 was 24 tonnes (Environment Canada, 1996a), although gasoline and diesel fuel contain little or no butadiene (US EPA, 1989).

CPPI (1997) estimated that releases into the Canadian environment in 1994 were 1191 tonnes from prescribed forest burning, 3706 tonnes from wood space heating, 11 tonnes from natural gas/oil space heating, and 1–9 tonnes from cigarettes.

Butadiene is produced during the combustion of organic matter in both natural processes and human activities. In addition, it is produced commercially for use in the chemical polymer industry.

Butadiene is purified by extraction from a crude petroleum butadiene stream. In 1994, there was one Canadian commercial producer of butadiene (located in Sarnia, Ontario), with a domestic production of 103.7 kilotonnes. Importation into Canada from the USA was 1.7 kilotonnes in 1994. The Canadian domestic use of butadiene in 1994 amounted to 105.4 kilotonnes (98.3 kilotonnes for total domestic demand and 7.1 for export sales) (Camford Information Services, 1995). In the USA, total production in 1993 was 1.4 billion kilo grams.³ According to data summarized in IARC (1999), in 1996, production of butadiene in China (Taiwan), France, Germany, Japan, the Republic of Korea, and the USA was 129, 344, 673, 1025, 601, and 1744 kilo tonnes, respectively.

The largest end use of butadiene in Canada is the production of polybutadiene rubber (51.4 kilotonnes; 52.3% of total Canadian consumption for 1994) (Camford Information Services, 1995). Other derivatives produced include styrene-butadiene lattices (31.0 kilo tonnes; 31.5% of total Canadian consumption for 1994), nitrile-butadiene rubbers (10.0 kilotonnes; 10.2% for 1994), acrylonitrile-butadiene-styrene terpolymer (3.4 kilotonnes; 3.5% for 1994), and specialty styrene- butadiene rubbers (2.5 kilotonnes; 2.5% of total Cana dian consumption for 1994).

Butadiene has a long history of use, notably related to production of polymers. Several industrial and commercial products are manufactured with it or may contain it as a component. Examples include tires, car sealants, plastic bottles and food wrap, epoxy resins, lubricating oils, hoses, drive belts, moulded rubber goods, adhesives, paint, latex foams for carpet backing or underpad, shoe soles, moulded toys/household goods, medical devices, and chewing gum (CEH-SRI Interna tional, 1994; OECD, 1996).

Since butadiene is released primarily to air, its fate in that medium is of primary importance. Butadiene is not expected to persist in air, since it oxidizes rapidly with several oxidant species. Destruction of atmospheric butadiene by the gas-phase reaction with photochemi cally produced hydroxyl radicals is expected to be the dominant photo-initiated pathway. Products that can be formed include formaldehyde, acrolein, and furan. Destruction by nitrate radicals is expected to be a sig nificant nighttime process in urban areas. Acrolein, trans-4-nitroxy-2-butenal, and 1-nitroxy-3-buten-2-one have been identified as products of this reaction. Reaction with ozone is also rapid but less important than reaction with hydroxyl radicals. The products of the reaction of butadiene with ozone are acrolein, formalde hyde, acetylene, ethylene, formic acid, formic anhydride, carbon monoxide, carbon dioxide, hydrogen gas, hydro peroxyl radical, hydroxyl radical, and 3,4-epoxy-1- butene (Atkinson et al., 1990; Howard et al., 1991; McKone et al., 1993; US EPA, 1993).

Average atmospheric half-lives for photo-oxidation of butadiene, based on measured as well as calculated data, range from 0.24 to 1.9 days (Darnell et al., 1976; Lyman et al., 1982; Atkinson et al., 1984; Becker et al., 1984; Klöpffer et al., 1988; Howard et al., 1991; Mackay et al., 1993). However, half-lives for butadiene in air can vary considerably under different conditions. Estimations for atmospheric residence time in several US cities ranged from 0.4 h under clear skies at night in the summer to 2000 h (83 days) under cloudy skies at night in the winter. Daytime residence times for different cities within a given season varied by factors of 2–3. Nighttime residence times varied by larger factors. The differences between summer and winter conditions were large at all sites, with winter residence times 10–30 times greater than summer residence times (US EPA, 1993). Because of the long residence times under some conditions, especially in winter under cloudy conditions, there is a possibility of day-to-day carryover. Nonetheless, given the generally short daytime residence times, the net atmospheric lifetime of butadiene is short, and there is generally limited potential for long-range transport of this compound.

It is predicted from its physical/chemical properties that when butadiene is released into air, almost all of it will exist in the vapour phase in the atmosphere (Eisenreich et al., 1981; Environment Canada, 1998). Wet and dry deposition are not expected to be important as transfer processes. Evaporation from rain may be rapid, and the compound is returned to the atmosphere relatively quickly unless it is leached into the soil.

Volatilization, biodegradation, and oxidation by singlet oxygen are the most prominent processes involved in determining the fate of butadiene in water. The estimated half-lives of butadiene by reaction in water range from 4.2 to 28 days (Howard et al., 1991; Mackay et al., 1993).

The processes that are most prominent in deter mining the environmental fate of butadiene in sediment are biotic and abiotic degradation. The modelled half- lives of butadiene by reaction in sediment range from 41.7 to 125 days (Mackay et al., 1993).

Based on its vapour pressure and its solubility, volatilization of butadiene from soil and other surfaces is expected to be significant. Butadiene’s organic carbon/ water partition coefficient indicates that it should not adsorb to soil particles to a great degree and would be considered moderately mobile (Kenaga, 1980; Swann et al., 1983). However, the rapid rate of volatilization and the potential for degradation in soil suggest that it is unlikely that butadiene will leach into groundwater. Based on modelling predictions, the half-life of butadi ene by reaction, given by Howard et al. (1991) and Mackay et al. (1993), ranges from 7 to 41.7 days.

There are no measured bioconcentration factors. Butadiene is metabolized by the mixed-function oxidase system in higher organisms, which contributes to the expected lack of accumulation by many organisms. Estimated bioconcentration factors for butadiene in fish have been reported to range from 4.6 to 19 (Lyman et al., 1982; OECD, 1996). Even though estimation methods likely overestimate the true bioconcentration potential for a readily metabolized substance, they indicate that butadiene is not expected to bioconcentrate in aquatic organisms or to biomagnify in the aquatic food chain.

There are no reported measurements of plant root bioconcentration in soils. However, McKone et al. (1993) estimated the uptake of butadiene by roots from soil solution to be 1.84 litres/kg, which is the ratio of butadiene concentration in root (mg/kg, fresh mass) to the concentration in soil solution (mg/litre). The partition coefficient of butadiene concentration in roots (mg/kg, fresh mass) to concentration in soil solids (mg/kg) was estimated to range from 0.32 to 15 (dimensionless).

The partition coefficient of butadiene concentration in whole plants (mg/kg, fresh mass) to its concentra tion in soil solids (mg/kg) was estimated to range from 0.1 to 2.9 (dimensionless). The steady-state plant/air partition coefficient for foliar uptake of butadiene in plant leaves was estimated to be 0.63 m³/kg. There are no reported bioaccumulation data for any terrestrial invertebrates.

Fugacity modelling was conducted to provide an overview of key reaction, intercompartment, and advection (movement out of a system) pathways for butadiene and of its overall distribution in the environment. A steady-state, non-equilibrium model (Level III fugacity modelling) was run using the methods developed by Mackay (1991) and Mackay and Paterson (1991). Assumptions, input parameters, and results are presented in Environment Canada (1998). Based on butadiene’s physical/chemical properties, Level III fugacity model ling predicts that:

• when butadiene is released into air, the distribution of mass is almost 100% in air, with very small amounts in soil and water;
• when butadiene is released into water, the distribu tion of mass is 99.0% in water, with small amounts in air;
• when butadiene is released into soil, the distribu tion of mass is 38.6% in soil, 59.3% in air, and 2.1% in water.

Modelling predictions do not purport to reflect actual expected measurements in the environment but rather indicate the broad characteristics of the fate of the substance in the environment and its general distribution between media. Thus, when butadiene is discharged into air or water, most of it is expected to be found in the medium receiving the discharge directly. For example, if butadiene is discharged into air, almost all of it will exist in the atmosphere, where it will react rapidly and will also be transported away. If butadiene is discharged to water, it will react in water, and some will also evaporate into air. If butadiene is discharged to soil, most will be present in air or soil, where it will react (Mackay et al., 1993; Environment Canada, 1998).

Data on environmental levels and human exposure from Canada, the source country of the national assess ment on which this CICAD is based, are presented here as a basis for the sample risk characterization. Patterns of exposure in other countries are expected to be similar, although quantitative values may vary.

6.1.1 Ambient air

Butadiene was detected (detection limit 0.05 µg/m³) in 7314 (or 80%) of 9168 24-h samples collected between 1989 and 1996 from 47 sites across Canada.⁴ The mean concentration in all samples was 0.3 µg/m³ (in the calculation of the mean, a value of one- half the detection limit was assumed for samples in which levels were below the detection limit), and the maximum concentration measured was 14.1 µg/m³. Concentrations of butadiene in ambient air correspond ing to the 50th and 95th percentiles were 0.21 and 1.0 µg/m³, respectively. Concentrations were generally higher in urban areas, with a mean exposure to 0.4 µg/m³ (95th percentile 1.3 µg/m³) estimated as a "reasonable worst-case scenario," based on data from four sites. Similar levels were measured in smaller surveys in Canada (Bell et al., 1991, 1993; Hamilton- Wentworth, 1997; Conor Pacific Environmental, 1998).⁵ In areas influenced by industrial point sources of butadiene, concentrations in air were greater, with maximum and mean levels of 28 and 0.62 µg/m³, respectively (95th percentile 6.4 µg/m³) being measured between 1 and 3 km from the source (MOEE, 1995).

Butadiene has also been detected in air in enclosed structures. Concentrations of butadiene between 4 and 49 µg/m³ were measured during the winter months of 1994–1995 in Canadian underground parking garages (Environment Canada, 1994) because of its presence in vehicle exhaust. Similarly, butadiene was frequently detected in samples from 10 parking structures in Cali fornia, with the maximum concentration being 28 µg/m³ (Wilson et al., 1991). Butadiene has also been detected in urban road tunnels during rush hours in Australia (mean concentration 28 µg/m³; Duffy & Nelson, 1996) and Sweden (mean concentrations 17 µg/m³ and 25 µg/m³ in two tunnels; Barrefors, 1996). Butadiene was measured at concentrations ranging from 0.2 to 28 µg/m³ in 96 of 97 5-min air samples collected from a pumping island at randomly identified self-service filling stations in California (Wilson et al., 1991).

6.1.2 Surface water

No data on concentrations of butadiene in Cana dian lake, river, estuarine, or marine waters were identi fied in the literature. Butadiene is being monitored in effluents discharged into the St. Clair River from the butadiene production plant in Sarnia, Ontario. It was detected only twice, at 2 and 5 µg/litre, in 2103 compo site samples of aqueous effluent taken every 4 h in 1996 (detection limit 1 µg/litre). In daily sampling of effluents from the four individual outfalls (detection limit 1 µg/litre in 736 samples and 50 µg/litre in 789 samples), buta diene was detected in only three samples, at concentra tions of 21, 80, and 130 µg/litre.⁶

6.1.3 Groundwater

Butadiene was detected but not quantified in a groundwater plume near a waste site in Quebec where refinery oil residues and a variety of organic chemicals had been dumped (Pakdel et al., 1992).

6.2.1 Indoor air

In available surveys in Canada, 1,3-butadiene was detected up to 6 times more frequently in indoor air in homes than in corresponding samples of outdoor air, with concentrations being up to 10-fold higher indoors than outdoors (Bell et al., 1993; Hamilton-Wentworth, 1997; Conor Pacific Environmental, 1998).⁷ Concentrations in air of indoor environments are highly variable and depend largely on individual activities and circum stances, including the use of consumer products (e.g., cigarettes), the infiltration of vehicle exhaust from nearby traffic and possibly from attached garages, and cooking activities involving heated fats and oils (see section 6.2.3). While data are inadequate to determine the relative contributions of each of these potential indoor sources, the highest concentrations of butadiene in indoor air in Canada have generally been detected in indoor environments contaminated with environmental tobacco smoke (ETS). In a survey of 94 homes across Canada, the mean level in "non-smoking" homes was <1 µg/m³ (data censored by considering levels to be one- half the detection limit in samples in which butadiene was not detected), compared with a mean of 2.5 µg/m³ (data censored) in homes where smoking was present (Conor Pacific Environmental, 1998). Similarly, mean concentrations in indoor air from "non-smoking" loca tions in Windsor, Ontario, ranged from 0.3 to 1.6 µg/m³, while mean levels in "smoking" locations ranged from 1.3 to 18.9 µg/m³. At non-residential indoor sampling sites in Windsor, the frequency of detection of butadiene was 75–100% where ETS was present (Bell et al., 1993).

6.2.2 Drinking-water

There are no data available concerning the pres ence of butadiene in drinking-water. In an investigation on whether the use of polybutylene pipe in water distri bution systems is likely to result in the contamination of drinking-water with butadiene, Cooper⁸ did not detect the substance in water from these types of pipes (no further information was presented in the secondary account [CARB, 1992] of this study).

6.2.3 Food

There are no data available concerning the presence or concentrations of butadiene in food in Canada. In the USA, the migration of butadiene from rubber-modified plastic containers to food was inves tigated by McNeal & Breder (1987). Butadiene was detected in some of the containers, but was generally not detected in the foods (detection limits 1–5 ng/g). Similarly, in the United Kingdom, butadiene was not detected (detection limit 0.2 ng/g) in five brands of soft margarine, although its presence was demonstrated (at concentrations ranging from <5 to 310 ng/g) in the plastic containers (Startin & Gilbert, 1984). Butadiene has been detected in the emissions from heated cooking oils, including Chinese rapeseed, peanut, soybean, and canola oils, at levels ranging from 23 to 504 µg/m³ (Pellizzari et al., 1995; Shields et al., 1995).

6.2.4 Consumer products

Data on emissions of butadiene from potential indoor sources such as styrene-butadiene rubber were not identified.

Butadiene has been detected in both mainstream smoke and sidestream smoke from cigarettes in Canada and the USA. For 18 brands of Canadian cigarettes, the mean butadiene content ranged from 14.3 to 59.5 µg/cig arette (overall mean concentration 30.0 µg/cigarette) in the mainstream smoke and from 281 to 656 µg/cigarette (overall mean concentration 375 µg/cigarette) in the sidestream smoke, according to "preliminary" data (Labstat, Inc., 1995). The US DHHS (1989) reported that the vapour phase of mainstream smoke of non-filter cigarettes contained butadiene at levels of 25–40 µg/cig arette. Brunnemann et al. (1989) measured butadiene levels ranging from 16 to 75 µg/cigarette in mainstream smoke from seven brands of cigarettes and levels ranging from 205 to 361 µg/cigarette in the sidestream smoke from six types of cigarettes. As discussed in section 6.2.1, the presence of ETS contributes to elevated levels of butadiene in indoor air.

6.2.5 Occupational exposure

Potential occupational exposure to butadiene can occur in petroleum refining and related operations, pro duction of butadiene monomer, production of butadiene- based polymers, or the manufacture of rubber and plas tics products (IARC, 1999). Arithmetic mean concentra tions in petroleum and petrochemical operations in several European countries ranged from 0.1 to 6.4 mg/m³ during 1984–1987 (IARC, 1999; European Chemicals Bureau, 2001). Based on occupational hygiene surveys of butadiene production facilities in the United King dom, personal airborne exposures are generally below a mean concentration of 5 ppm (11 mg/m³), with most below 1 ppm (2.2 mg/m³). In polymer manufacture in the United Kingdom, most time-weighted average exposures are below 2–3 ppm (4.4–6.6 mg/m³). Similar concentra tions were reported in other facilities in the European Union (IARC, 1999). In monomer production facilities in the USA surveyed in 1985, arithmetic mean concen trations ranged from 1 to 277 mg/m³, while those in polymer production industries ranged from 0.04 to 32 mg/m³ (IARC, 1999).

The database on the toxicokinetics and metabolism of butadiene is relatively extensive. The proposed metabolism is outlined in Figure 1, based on the path ways described by Henderson et al. (1993, 1996) and Himmelstein et al. (1997). Available data for the path ways most extensively investigated indicate that metabo lism is qualitatively similar among the various species studied, although there may be quantitative differences in the amount of butadiene absorbed as well as in metabolic rates and the proportion of metabolites generated. These differences appear to be in concordance with the observed variation in sensitivity to butadiene-induced toxic effects of the few strains of rodent species tested to date, in that mice appear to metabolize a greater propor tion of butadiene to active epoxide metabolites than do rats. While less of these metabolites are also formed in samples of human tissues in vitro than in those of mice, available data are insufficient to characterize interindi vidual variability in humans. Although there are known genetic polymorphisms for a number of the enzymes involved in the metabolism of butadiene, information on genotype was not included in most investigations in humans.

Based on the metabolic pathways described in Figure 1, butadiene is first oxidized via cytochrome P- 450 enzymes (primarily P-450 2E1 in humans, although other isoforms may also be involved, the relative contrib utions of which vary between tissues and species) to the monoepoxide 1,2-epoxy-3-butene, or EB, which is sub sequently further oxidized via P-450 enzymes to the diepoxide 1,2,3,4-diepoxybutane, or DEB, or hydrolysed via epoxide hydrolase (EH) to butenediol (1,2-dihy droxy-3-butene). The monoepoxide, the diepoxide, and the butenediol may all be conjugated with glutathione (GSH) to form mercapturic acids, which are eventually eliminated in the urine. Hydrolysis of the diepoxide via epoxide hydrolase or oxidation of the butenediol via cytochrome P-450 will result in the formation of the monoepoxide diol (EBdiol). A small amount of buta diene may be converted to 3-butenal, which is subse quently transformed to crotonaldehyde (about 2–5% of the amount that is oxidized to the monoepoxide in human liver microsomes [Duescher & Elfarra, 1994] or microsomes of kidney, lung, or liver of B6C3F₁ mice [Sharer et al., 1992]). However, this pathway has not been extensively investigated, nor was crotonaldehyde detected in a sensitive analysis (using nuclear magnetic resonance spectroscopy) of urinary metabolites of rats and mice exposed to ¹³C-butadiene (Nauhaus et al., 1996).

Metabolism of butadiene and subsequent conver sion of EB to DEB may also take place to a more limited degree in the bone marrow (e.g., Maniglier-Poulet et al., 1995) by means other than P-450 oxidation (possibly via myeloperoxidase; Elfarra et al., 1996), based on in vitro observations and the detection of the epoxides in the bone marrow of rodents (Thornton-Manning et al., 1995a, 1995b), although this potential pathway has not yet been extensively investigated. EB may also react with both myeloperoxidase and chloride to form a chlorohydrin (1-chloro-2-hydroxy-3-butene) (Duescher & Elfarra, 1992). Metabolites arising from other possible pathways have been identified in the urine of mice exposed to butadiene (including metabolites known to be derived from metabolism of acrolein or acrylic acid) (Nauhaus et al., 1996), but no further research has yet been generated.

There is a substantial amount of evidence from in vitro and in vivo investigations that B6C3F₁ mice oxidize butadiene to the monoepoxide via P-450 in the liver to a greater extent than do Sprague-Dawley rats and humans. Levels of EB in the blood and other tissues of mice were two- to eightfold higher than those in rats exposed to similar levels of butadiene (Bond et al., 1986; Himmel stein et al., 1994, 1995; Bechtold et al., 1995; Thornton- Manning et al., 1997).

Available data also suggest that there are similar species differences in the amount of the diepoxide formed from oxidation of the monoepoxide. Levels of DEB were 40- to 160-fold higher in blood and other tissues of B6C3F₁ mice than in Sprague-Dawley rats exposed to the same concentration of butadiene (Thorn ton-Manning et al., 1995a, 1995b). While concentrations of EB at various sites were similar in male and female rats, levels of DEB were at least fivefold higher in females than in males, which correlates with the greater incidence of tumours in female rats. Although the mam mary gland is a target tissue in rats, extended exposure to butadiene at 8000 ppm (17 696 mg/m³) for 10 days did not result in any accumulation of DEB at this site (Thornton-Manning et al., 1998), which suggests that DEB may not play a significant role in the induction of mammary tumours in rats. Available in vitro data in human liver and lung samples suggest that humans also form less of the active metabolites of butadiene than do mice (although somewhat varying results have been reported with respect to the magnitude of the differences between species) (Csanády et al., 1992; Duescher & Elfarra, 1994; Krause & Elfarra, 1997).

Although epoxide metabolites of butadiene are formed to a greater extent in mice than in rats or humans, they are also cleared via glutathione conjugation more rapidly in mice (Kreuzer et al., 1991; Sharer et al., 1992; Boogaard et al., 1996a, 1996b). Conversely, hydrolysis of EB and DEB is greater in humans than in rats (based on in vitro data, as DEB has not been detected in tissues of exposed humans), and hydrolysis of EB and DEB in rats is in turn greater than that in mice (Csanády et al., 1992; Krause et al., 1997). In both humans and monkeys, removal of EB via hydrolysis appears to predominate over conjugation with glutathione, based on analysis of urinary metabolites (Sabourin et al., 1992; Bechtold et al., 1994). Although hydrolysis of the epoxide metabo lites is generally considered to be a detoxifying mecha nism, it may also lead to the formation of the diolepox ide, EBdiol, which is biologically reactive. However, no data were identified on species differences in the forma tion of EBdiol via metabolism of both epoxide metabo lites.

The formation of stable adducts of both the mono epoxide and monoepoxide diol metabolites of butadiene with the N-terminal valine of haemoglobin has been observed in experimental animals and humans exposed to butadiene (Albrecht et al., 1993; Osterman-Golkar et al., 1993, 1996; Neumann et al., 1995; Sorsa et al., 1996b; Tretyakova et al., 1996; Pérez et al., 1997).⁹ Consistent with the greater formation of epoxide metab olites, greater concentrations of haemoglobin–EB adducts were measured in mice than in rats exposed to the same concentration of butadiene. However, levels of haemoglobin–EB adducts in butadiene-exposed workers, although significantly elevated compared with levels in non-exposed workers, were considerably less than would be expected on the basis of results of studies in mice and rats (Osterman-Golkar et al., 1993). Based on observa tions in rats and humans exposed to butadiene, levels of haemoglobin–EBdiol adducts are substantially greater than levels of haemoglobin–EB adducts (although it is noted that the same adduct can result from binding with DEB). Metabolites of butadiene may also form adducts with DNA (see sections 8.5 and 9.2.3).

In addition to quantitative interspecies differences in the metabolism of butadiene, there is also evidence that there is significant variation within the human popu lation. Indeed, although available data are inadequate to assess interindividual variation in metabolism, which has been observed in in vitro investigations in microsomes from a small number of subjects (Boogaard & Bond, 1996; Krause et al., 1997), there has been significant interindividual variability in the extent of formation of haemoglobin adducts with butadiene metabolites in human populations (Neumann et al., 1995; Osterman- Golkar et al., 1996). Such variability is not unexpected, in view of the complexity of the metabolic pathways involved in the biotransformation of butadiene: i.e., the three principal enzymatic processes that determine the extent of exposure to the putatively toxic epoxide metabolites, namely formation via cytochrome P-450 2E1 and removal via epoxide hydrolase and glutathione conjugation. For example, the inducibility of cytochrome P-450 2E1 by low molecular weight compounds such as ethanol is likely to contribute to interindividual variabil ity in sensitivity. Moreover, genetic polymorphisms for glutathione-S-transferases and epoxide hydrolase might also contribute to considerable variation in sensitivity. While the influence of genotype for epoxide hydrolase has not been well investigated (although data indicate that hydrolysis of EB predominates over oxidation and glutathione conjugation in humans), interindividual sensitivity to the genetic effects of the epoxide metabo lites in in vitro studies has been clearly related to geno type for the glutathione-S-transferases (see section 9.2.3).

Although few data are available, butadiene appears to be of low acute toxicity in experimental animals, with reported LC₅₀ values for rats and mice of >100 000 ppm (>221 000 mg/m³). Lowest LC₅₀ values for butadiene are reported for mice, at 117 000 ppm (256 000 mg/m³) (duration not specified) (Batinka, 1966) and 121 000 ppm (268 000 mg/m³) (2 h) (Shugaev, 1969). The nervous system and the blood appear to be the principal targets; however, in only one study were data sufficient to determine a lowest-observed-effect level (LOEL) of 200 ppm (442 mg/m³) for haematological effects (Leavens et al., 1997). Exposure to butadiene for 7 h caused a concentration-dependent depletion (by as much as 80%) of cellular non-protein sulfhydryl content of liver, lung, or heart in mice, with a LOEL of 100 ppm (221 mg/m³) (Deutschmann & Laib, 1989). Depletion of non-protein sulfhydryl content may inhibit detoxification of epoxide metabolites vie glutathione conjugation.

No investigations in experimental animals on the potential for irritation or sensitization of butadiene have been identified.

The majority of short-term and subchronic studies were designed as either range-finding studies preliminary to chronic bioassays or investigations of potential mech anisms of action for butadiene-induced cancer and are not adequate for determination of critical effect levels. Effects on body weight were observed in B6C3F₁ mice exposed to 625 ppm (1383 mg/m³) butadiene or more for 2 weeks; no histopathological changes were noted at any concentration at or below 8000 ppm (17 696 mg/m³) (NTP, 1984).

Haematological effects consistent with megalo blastic anaemia and effects on bone marrow, including alterations in stem cell development, have been observed in two strains of mice (B6C3F₁ and NIH Swiss) exposed to 1000 or 1250 ppm (2212 or 2765 mg/m³) butadiene for up to 31 weeks (Irons et al., 1986a, 1986b; Leider man et al., 1986; Bevan et al., 1996). Other effects, including decreased survival and body weight gain (with males being more sensitive than females), altered organ weights, and ovarian or testicular atrophy, have also been observed in B6C3F₁ mice exposed subchronically to similar or higher levels of butadiene (NTP, 1984; Bevan et al., 1996). In addition, an increased incidence of a variety of tumours has been observed in B6C3F₁ mice exposed to 625 ppm (1383 mg/m³) butadiene for as little as 13 weeks (NTP, 1993) (see section 8.4). Although histopathological changes and haematological effects were reported in early studies in rats exposed to low concentrations (3 or 10 mg/m³) (Batinka, 1966; Ripp, 1967; Nikiforova et al., 1969), these results were not confirmed in more recent investigations of rats exposed for up to 13 weeks to much higher concentra tions (e.g., 17 600 mg/m³) (e.g., Crouch et al., 1979; Bevan et al., 1996). In view of the limitations of the studies in rats, it is not possible to draw any conclusions regarding species differences in response to subchronic exposure to butadiene.

The carcinogenic potential of inhaled butadiene has been studied in two strains of mice and one strain of rats. Butadiene was a multi-site carcinogen in all identified long-term experiments, inducing common and rare tumours in mice and rats, although there appear to be marked species and strain differences in sensitivity. In an early bioassay by the National Toxicology Program (NTP, 1984) in which male and female B6C3F₁ mice were exposed to 0, 625, or 1250 ppm (0, 1383, or 2765 mg/m³) butadiene for up to 61 weeks, there were exposure-related increases in the incidences of malignant lymphomas, cardiac haemangiosarcomas (an extremely rare tumour in B6C3F₁ mice), and lung tumours in both sexes. There were also increased incidences of papillo mas or carcinomas of the forestomach, hepatocellular adenomas or carcinomas, ovarian granulosa cell tumours, acinar cell carcinomas of the mammary gland, brain gliomas, and Zymbal gland carcinomas (the latter two tumour types have only rarely been observed in this strain of mice at the NTP) in one or both sexes.

Because of the poor survival of mice in the earlier bioassay and to better characterize the exposure– response relationship, the NTP (1993) subsequently exposed B6C3F₁ mice to lower concentrations (0, 6.25, 20, 62.5, 200, or 625 ppm [0, 13.8, 44.2, 138, 442, or 1383 mg/m³]) of butadiene for up to 2 years. Survival was again decreased in most groups (³ 20 ppm [³ 44.2 mg/m³]); at the highest concentration, deaths were principally due to lymphatic lymphomas, which appeared to arise from the thymus and occurred as early as week 23. Non-neoplastic effects observed in exposed mice included a variety of haematological effects, alter ations in organ weights, bone marrow atrophy and hyper plasia, atrophy of the thymus, atrophy and angiectasis of the ovaries, uterine atrophy, mineralization of the cardiac endothelium, liver necrosis, and olfactory epithelial atrophy. There were significant increases in the inci dence of tumours at a variety of sites (incidence data presented in Table 2), including malignant lymphomas (particularly lymphocytic lymphomas), histiocytic sarco mas, cardiac haemangiosarcomas, Harderian gland adenomas and carcinomas, hepatocellular adenomas and carcinomas, alveolar/bronchiolar adenomas and carcino mas, mammary gland adenoacanthomas, carcinomas, and malignant mixed tumours, ovarian granulosa cell tumours, and forestomach squamous cell papillomas and carcinomas, particularly when the incidences were adjusted for survival. The incidence of alveolar/bronchi olar adenomas or carcinomas was significantly increased in females at all concentrations (i.e., ³ 6.25 ppm [³ 13.8 mg/m³]). Low incidences of uncommon tumours, such as preputial gland carcinomas, Zymbal gland car cinomas in males, and renal tubule adenomas in both sexes, were also suspected of being related to exposure. In addition, exposure to butadiene induced malignant tumours at several sites, whereas, in general, tumours at the same sites in control animals were benign.

The NTP also conducted a "stop-exposure" experiment in male B6C3F₁ mice designed to investigate whether tumour induction was associated with the exposure concentration or the duration of exposure. Animals were exposed to 200 ppm (442 mg/m³) for 40 weeks or 625 ppm (1383 mg/m³) for 13 weeks (both equivalent to 8000 ppm-weeks) or to 312 ppm (690 mg/m³) for 52 weeks or 625 ppm (1383 mg/m³) for 26 weeks (both equivalent to 16 000 ppm-weeks); all groups were observed for the remainder of the 2-year study. Again, survival was reduced in all exposed mice, largely due to malignant neoplasms, with significant increases in the incidences of lymphocytic lymphomas, histiocytic sarcomas, cardiac haemangiosarcomas, Harderian gland adenomas or carcinomas, hepatocellular adenomas or carcinomas, alveolar/bronchiolar adenomas or carcinomas, and squamous cell papillomas or carcino mas of the forestomach (even in mice exposed for only 13 weeks) (incidence data presented in Table 2). In addition, low incidences of several uncommon tumour types (preputial gland carcinomas, Zymbal gland car cinomas, malignant gliomas and neuroblastomas of the brain, Harderian gland carcinomas, and renal tubule adenomas) were again observed in one or more of the exposed groups. Concentration may be more important than the duration of exposure in tumour development, as the incidence of malignant lymphomas and squamous cell carcinomas of the forestomach was greater in the groups that had been exposed to 625 ppm (1383 mg/m³) for a shorter period than in those exposed to 200 ppm (442 mg/m³) for a longer period (i.e., similar total cumu lative exposure) (NTP, 1993).

Acute exposure of B6C3F₁ mice for 2 h to up to 10 000 ppm (22 120 mg/m³) butadiene, followed by observation for 2 years, did not induce an increased incidence of tumours at any site (Bucher et al., 1993).

Sensitivity to butadiene-induced thymic lympho ma/leukaemia appears to be enhanced by the presence of an endogenous ecotropic retrovirus in B6C3F₁ mice, as the incidence of this tumour was greater in male B6C3F₁ mice exposed to 1250 ppm (2765 mg/m³) butadiene for 52 weeks than in male Swiss mice, which do not express an endogenous retrovirus (57% versus 14%). Exposed mice of both strains had elevated incidences of thymic lymphoma/leukaemia compared with controls, as did B6C3F₁ mice exposed to 1250 ppm (2765 mg/m³) for 12 weeks and then observed for an additional 40 weeks, although the MuLV env sequence for the retrovirus was detected only in tumours of the B6C3F₁ mice. Other tumours reported in the mice exposed for 52 weeks included haemangiosarcomas of the heart (mainly in B6C3F₁ mice) and lung tumours. Neoplasms of the glandular and non-glandular stomach were observed in the B6C3F₁ mice, whereas adenocarcinomas of the Harderian gland and the thyroglossal duct were observed in the Swiss mice (Irons et al., 1989).

In the only identified long-term bioassay in rats (Hazleton Laboratories Europe Ltd., 1981a; Owen et al., 1987; Owen & Glaister, 1990), male and female Sprague-Dawley rats were exposed to 0, 1000, or 8000 ppm (0, 2212, or 17 696 mg/m³) butadiene for up to 111 weeks. At 8000 ppm (17 696 mg/m³), survival was reduced in both sexes; there were also changes in the relative weights of a number of organs in males at this concentration, along with an increase in the severity of nephrosis of the kidney relative to controls. Relative liver weights were increased in all exposed groups, although there were no exposure-related histopathologi cal effects on the liver. At 8000 ppm (17 696 mg/m³), there were increased incidences of follicular cell adeno mas and carcinomas of the thyroid gland in females and exocrine adenomas of the pancreas in males (with a carcinoma occurring in a rat of either sex) (incidence data presented in Table 3). In females, the incidence of benign or malignant mammary gland tumours, along with the incidence of animals with multiple mammary gland tumours, was increased at both 1000 and 8000 ppm (2212 and 17 696 mg/m³). The incidence of sarcomas of the uterus and carcinomas of the Zymbal gland increased significantly with level of exposure in females; in addi tion, a Zymbal gland carcinoma occurred in one male rat at each exposure level. The incidence of Leydig cell tumours of the testes was increased in both groups of exposed males. The investigators suggested that the occurrence of tumours of the testes and Zymbal gland may have been unrelated to exposure, as the incidences observed were reportedly similar to those in other control rats of the same strain in the study laboratory, although it is noted that Zymbal gland tumours were noted in the chronic bioassays in mice discussed above.

Both the mono- and diepoxide metabolites (EB and DEB) have induced local tumours at the site of applica tion in Swiss mice or Sprague-Dawley rats (Van Duuren et al., 1963, 1965, 1966), although available studies are inadequate to evaluate species differences in sensitivity.

It has been hypothesized that the observed greater sensitivity of B6C3F₁ mice compared with Sprague- Dawley rats to the induction of thymic lymphoma by butadiene may be related to differences in the potential of EB to affect haematopoietic stem cell differentiation observed in in vitro investigations, as suppression of clonogenic response was greater in bone marrow cells from C56BL/6 mice than in those from Sprague-Dawley rats or humans; it was also hypothesized that the sub population of progenitor cells affected in mice is not present in humans (Irons et al., 1995).

The genotoxicity of butadiene has been investigated in a limited range of in vitro assays and a more extensive range of in vivo tests. Butadiene was mutagenic in Salmonella typhimurium strains TA1530 and TA1535 in the presence of metabolic activation with rodent or human S9 preparations (de Meester et al., 1978, 1980; Arce et al., 1990; NTP, 1993; Araki et al., 1994), although it was generally inactive in strains TA97, TA98, and TA100 with or without exogenous activation under similar experimental conditions (Victorin & Ståhlberg, 1988; Arce et al., 1990; NTP, 1993). Results of mouse lymphoma assays have been conflicting, with an increased frequency of mutations at the tk locus in one study at very high concentrations (i.e., 200 000–800 000 ppm [442 400–1 796 600 mg/m³]) in the presence of metabolic activation (Sernau et al., 1986), while there was no convincing activity at concen trations of up to 300 000 ppm (663 600 mg/m³) in another study (although the authors noted that the lack of a positive response may have been due to the low solubility of butadiene in the culture medium; NTP, 1993). Butadiene dissolved in ethanol induced sister chromatid exchanges in cultured mammalian cells (hamsters and humans) (Sasiadek et al., 1991a, 1991b), while in vitro exposure to gaseous butadiene did not induce this effect in preparations from rats, mice, and humans (Arce et al., 1990; Walles et al., 1995).

An overview of the results of available in vivo assays for genotoxicity in germ and somatic cells in mice and rats is presented in Table 4; in general, the data are consistent with species-specific differences in sensitivity to butadiene-induced genetic damage, likely related to the quantitative differences in the formation of active metabolites, although fewer studies have been conducted in rats. Butadiene induced dominant lethal mutations in two strains of mice (CD-1 and (102/E1 × C3H/E1)F₁) following short-term or subchronic exposure of males to concentrations as low as 500 ppm (1106 mg/m³) for 5 days or 65 ppm (144 mg/m³) for 4 weeks; however, exposure to 6250 ppm (13 825 mg/m³) for 6 h did not induce dominant lethal mutations in CD-1 mice. The results of these studies, which depended upon the timing of mating relative to exposure, suggested that the induc tion of dominant lethal mutations in mice was likely caused by effects on mature germ cells. In the only simi lar study in rats identified, there was no evidence of dominant lethal mutations in Sprague-Dawley rats exposed to up to 1250 ppm (2765 mg/m³) butadiene for 10 weeks.

Short-term exposure to 500 or 1300 ppm (1106 or 2876 mg/m³) butadiene also induced an exposure-related increase in the incidence of heritable chromosomal

translocations in mice; an increased incidence of chromosomal aberrations was also noted in zygotes of male mice exposed to ³ 500 ppm (³ 1106 mg/m³) for 5 days. Other butadiene-induced effects observed in male germ cells of mice include sperm head abnormal ities, micronuclei in spermatids, and DNA damage (strand breaks and alkaline-labile sites). Investigations of these end-points in rats have not been identified.

Butadiene was consistently genotoxic in somatic cells of several strains of mice, inducing chromosomal aberrations, sister chromatid exchanges, and micronuclei in numerous assays; micronuclei have been observed following exposure to concentrations as low as 6.25 ppm (13.8 mg/m³) butadiene for 13 weeks or 62.5 ppm (138 mg/m³) for 8 h. Although only few studies were identified, these effects were not observed in rats exposed to much higher concentrations. However, gene mutations at the hprt locus have been induced in both mice and rats, with a four- to sevenfold greater muta genic potency being determined for mice than for rats. Mutagenic activity was also observed in two transgenic mouse systems and in the mouse spot test. Binding to DNA has been observed in all strains of mice and rats tested; following exposure to butadiene, adducts of both guanine and adenine with the monoepoxide as well as the monoepoxide diol metabolites (EB and EBdiol, respec tively) have been observed. The degree of adduct forma tion was generally similar in the two species or, in some studies, up to twofold greater in mice than in rats. Similarly, there was little quantitative difference in the amount of butadiene-induced single strand breaks in DNA of mice and rats. DNA–DNA and DNA–protein cross-links were noted in one of two studies in mice, but not in rats exposed to higher concentrations of butadiene.

Metabolites of butadiene have also been mutagenic and clastogenic in numerous in vitro and in vivo assays (see Table 4 for overview of results of in vivo assays). EB, DEB, and EBdiol all induced mutations in bacteria and yeast in the absence of exogenous metabolic activa tion (IARC, 1992; NTP, 1993; Thier et al., 1994; Adler et al., 1997); mutagenic activity was also observed for all three metabolites at two foci in human TK6 lymphoblas toid cells, with DEB being much more potent (Cochrane & Skopek, 1993, 1994a). Conversely, the monoepoxide was much more potent than the diepoxide in the induc tion of mutations at the lacI transgene of fibroblasts obtained from a transgenic rat strain (Saranko & Recio, 1998; Saranko et al., 1998). Both EB and DEB also induced sister chromatid exchanges, chromosomal aberrations, and micronuclei in cultured mammalian (including human) cells (IARC, 1992; Xi et al., 1997). Aneuploidy in chromosomes 12 and X was also induced in human lymphocytes, which is notable in view of the fact that aneuploidy in these chromosomes is commonly observed in lymphocytic leukaemias (Xi et al., 1997). In vitro exposure to DEB, but not EB or EBdiol, induced micronuclei in spermatids isolated from rats (Sjöblom & Lähdetie, 1996).

The monoepoxide, diepoxide, and monoepoxide diol metabolites all induced micronuclei in germ cells of male mice and rats; in one of these studies, the magni tude of the effect was greater in Lewis rats than in F₁(102 × C3H) mice. There were no consistent patterns in the relative potency of the three metabolites. Chromosomal aberrations in zygotes produced by exposed males and dominant lethal mutations were induced by DEB in mice (strains (C57Bl/Cne × C3H/Cne)F₁ and (102/E1 × C3H/ E1)F₁), respectively), whereas EB and EBdiol did not induce dominant lethal mutations. In the only identified investigation of the potential effects on female germ cells, pre-mating exposure of female B6C3F₁ mice to DEB resulted in an increased frequency of chromosomal aberrations in embryos in the absence of ovarian toxicity.

EB, DEB, and EBdiol were also genotoxic in somatic cells (bone marrow, peripheral blood, lung, and spleen), inducing sister chromatid exchanges, chromo somal aberrations, or micronuclei in several strains of mice, rats, and hamsters, with little consistent evidence of interspecies differences in sensitivity; in general, the diepoxide was more potent than the monoepoxide or the monoepoxide diol. Although negative results were obtained in Lewis rats, both EB and DEB induced an increased frequency of hprt^– mutations in B6C3F₁ mice and F344 rats, with rats being more sensitive than mice, which may be related to slower clearance in rats. EB induced mutations in the bone marrow of lacI transgenic rats, but not in lacI transgenic mice; DEB did not ind