Butadiene

1.1.2. Structural and molecular formulae and relative molecular mass

Diepoxybutane

Diepoxybutane is the racemic mixture of four different isomers, with the following Chem. Abstr. Serv. Reg. Nos: 1464-53-5, diepoxybutane; 298-18-0, (±)-diepoxybutane; 564-00-1, meso-diepoxybutane; 30419-67-1, d-diepoxybutane; 30031-64-2, l-diepoxy-butane.

1.1.2. Structural and molecular formulae and relative molecular mass

1.1.5. Analysis

Selected methods for the analysis of butadiene in various matrices are listed in Table 1. Those for the analysis of butadiene in air have been evaluated; there appears to be no single preferred method, but more recent ones give a higher performance. Thermal desorption provides high levels of accuracy and precision (Bianchi et al., 1997).

Table 1

Selected methods for the analysis of butadiene in various matrices.

Several gas detector tubes are used in conjunction with common colorimetric reactions to detect butadiene. The reactions include the reduction of chromate or dichromate to chromous ion and the reduction of ammonium molybdate and palladium sulfate to molybdenum blue (Saltzman & Harman, 1989).

Passive dosimeters that use different techniques (thermal desorption and gas chromatography, colorimetric reactions) are also available for the detection of butadiene.

1.2.1. Production

Butadiene was first produced in the late nineteenth century by pyrolysis of various organic materials. Commercial production began in the 1930s (Sun & Wristers, 2002).

(c) Production volume

An estimated 9.3 million tonnes of butadiene were produced worldwide in 2005 (CMAI, 2006). Production volumes for different regions for the years 2004 and 2006 are given in Table 2.

Table 2

Butadiene production (in tonnes) by world region from 1981 to 2006.

World capacity grew by 3.5% per year between 1997 and 2002. During that period, most of the increase in capacity occurred in Asia, South America and the Middle East. Asia is now the largest producer of butadiene, and accounts for one-third of the world capacity (Walther, 2003).

Diepoxybutane is not believed to be produced commercially except in small quantities for research purposes (National Library of Medicine, 2008).

1.2.2. Use

Butadiene is used primarily in the production of synthetic rubbers and polymers. These polymers are used in a wide variety of industrial and consumer products, to improve their functionality, performance and safety and lower their costs. Butadiene-based products are important components of automobiles, construction materials, appliance parts, computers and telecommunications equipment, clothing, protective clothing, packaging and household articles (White, 2007).

The synthetic rubbers that are produced from butadiene include styrene–butadiene rubber, polybutadiene rubber, styrene–butadiene latex, chloroprene rubber and nitrile rubber. Important plastics that contain butadiene as a monomeric component are shock-resistant polystyrene, a two-phase system that consists of polystyrene and polybutadiene; polymers that consist of acrylonitrile, butadiene and styrene; and a copolymer of methyl methacrylate, butadiene and styrene, which is used as a modifier for poly(vinyl)chloride. Butadiene is also used as an intermediate in the production of chloroprene, adiponitrile and other basic petrochemicals (White, 2007).

Diepoxybutane has been proposed for use in curing polymers and cross-linking textile fibres (National Library of Medicine, 2008).

1.3.1. Natural occurrence

Butadiene is not known to occur as a natural product.

1.3.2. Occupational exposure

According to the 1990–93 CAREX database (see General Remarks) for 15 countries of the European Union (Kauppinen et al., 2000) and the 1981–83 US National Occupational Exposure Survey (NOES, 1997), approximately 31 500 workers in Europe and 50 000 workers in the USA were potentially exposed to butadiene.

Based on data from CAREX, the major categories of industrial exposure to butadiene in 15 European countries are the manufacture of industrial chemicals (8000 persons), rubber products (7000 persons), plastic products (7000 persons), petroleum refining (2200 persons) and building construction (1600 persons) (Kauppinen et al., 2000).

In the studies presented below, the accuracy of the levels of exposure to butadiene measured with the methods used until the mid-1980s may have been affected by the inability to distinguish between butadiene and other C₄ compounds, low desorption efficiency at low concentrations, possible sample breakthrough in charcoal tubes and possible loss during storage (Lunsford et al., 1990; Bianchi et al., 1997).

(a) Petroleum refining and butadiene monomer production

Detailed industrial hygiene surveys were conducted in the USA by the National Institute for Occupational Safety and Health in 1985 in four of 10 facilities where butadiene was produced by solvent extraction of C₄ fractions that originated from ethylene co-product streams (Krishnan et al., 1987). Levels of butadiene to which workers in various job categories were exposed are summarized in Table 3. Jobs that required workers to handle or transport containers, such as emptying sample cylinders or loading and unloading tank trucks or rail cars, presented the greatest potential exposure. Geometric means of full-shift exposure levels for other job categories were below 1 ppm [2.2 mg/m³]. Short-term samples showed that activities such as open-loop sampling and emptying cylinders were associated with peak exposures of 100 ppm [220 mg/m³]. Full-shift area samples indicated that ambient concentrations of butadiene were greatest in the rail car terminals (geometric mean, 1.8 ppm [3.9 mg/m³]) and in the tank storage farm (2.1 ppm [4.7 mg/m³]).

Table 3

Eight-hour time-weighted average exposure levels in personal breathing zone samples at four butadiene monomer production facilities in the USA, 1985.

Monitoring in a plant in Finland generally indicated ambient air levels of butadiene of less than 10 ppm [22 mg/m³] at different sites (33 samples; mean sampling time, 5.3 h). In personal samples for 16 process workers, the concentrations ranged from < 0.1 to 477 ppm [< 0.22–1050 mg/m³] (mean, 11.5 ppm [25 mg/m³]; median, < 0.1 ppm [< 0.22 mg/m³]; 46 samples; mean sampling time, 2.5 h). The highest concentrations were measured during the collection of samples, for which protective clothing and respirators were used (Work Environment Fund, 1991).

A study of biological monitoring for the mutagenic effects of exposure to butadiene reported estimated average exposures of 1 ppm [2.2 mg/m³] for workers in a butadiene monomer plant. Ambient air concentrations in production areas averaged 3.5 ppm [7.7 mg/m³], while average concentrations of 0.03 ppm [0.07 mg/m³] were reported for the control area (Ward, J.B. et al., 1996).

Levels of exposure to butadiene of workers in various job groups in the production and distribution of gasoline (see IARC, 1989) are shown in Table 4. Table 5 shows exposures in 1984–87 of workers in different areas of petroleum refineries and petrochemical facilities where crude butadiene was produced (usually a C₄ stream obtained as a by-product of ethylene production). Table 6 shows more recent data on crackers at butadiene production plants for the years 1986–93 (ECETOC, 1997).

Table 4

Personal exposures to butadiene associated with gasoline in 1984–85 in 13 European countries (540 measurements).

Table 5

Eight-hour time-weighted average concentrations of butadiene to which workers in different jobs in petroleum refineries and petrochemical facilities were exposed in the USA, 1984–87.

Table 6

Personal exposures to butadiene of crackers in butadiene production plants in the European Union.

Exposure data from 15 monomer extraction sites for the years 1984–93 showed that less than 10% of the measured concentrations exceeded 5 ppm [11 mg/m³] (Table 7); in 1995 (Table 8), personal exposure levels in general were below 5 ppm [11 mg/m³] (ECETOC, 1997).

Table 7

Personal exposures to butadiene in extraction units of butadiene production plants in the European Union.

Table 8

Personal exposures to butadiene at 15 monomer extraction sites in the European Union in 1995.

In 1998, personal exposure to butadiene was measured for 24 workers in a monomer production facility in the Czech Republic. The mean (± standard deviation [SD]) concentration of butadiene, calculated from 217 individual time-weighted average (TWA) measurements, was 0.6 ± 2.1 mg/m³ [0.27 ± 0.95 ppm]. The personal TWA measurements from all monomer production workers ranged from undetectable to 19.9 mg/m³. The mean concentration for the control group was 0.03 ± 0.03 mg/m³ [0.01 ± 0.01 ppm], calculated from 28 personal TWA exposure measurements (Albertini et al., 2003a).

Personal exposure to butadiene of 10 workers who held different jobs in a petrochemical plant in Finland was assessed using passive monitors shortly after the threshold limit value (TLV) of 1 ppm had come into force. A total of 119 personal breathing zone samples were taken and 117 were analysed. Of these, 32 (27%) samples were under the limit of quantification (0.029 mg/m³ [0.013 ppm] in a 20.5-L sample), 81 samples (69%) were between the limit of quantification and 1 ppm [2.2 mg/m³] and four samples (3%) were over the Finnish occupational exposure limit of 1 ppm. The mean value of all samples was 0.17 ppm [0.38 mg/m³] and the mean value of the samples that exceeded the Finnish occupational exposure limit was 1.75 ppm [3.87 mg/m³]. The mean level of exposure varied significantly (p = 0.03) between the 10 workers. Smoking did not significantly affect the values, but the seasonal effect was significant (p = 0.02) (Anttinen-Klemetti et al., 2004).

The occupational exposure of 42 workers in a petrochemical plant in Italy where butadiene was produced and used to prepare polymers was assessed by biomonitoring. The control group originated from the same industrial complex and included 43 workers who had no significant occupational exposure to butadiene. Active sampling from the breathing zone of the workers was performed during a full shift. Each exposed worker was assessed three to four times over a period of 6 weeks during different shifts. The mean exposure level of the control group was 0.9 µg/m³ [0.4 ppb] (SD, 1.0) and the lowest and highest values were < 0.1 and 3.8 µg/m³ [< 0.05 and 1.7 ppb], respectively. The mean exposure level of the exposed group was 11.5 mg/m³ [5.2 ppb] (SD, 35.8) and the lowest and highest values were < 0.1 and 220.6 µg/m³ [< 0.04 and 99.8 ppb], respectively (Fustinoni et al., 2004).

An exposure assessment was carried out in southern Taiwan, China, on a 120-acre [486 000 m²] petrochemical complex that comprised 11 different manufacturing plants. Butadiene was produced in two of the plants, which had an annual production of about 156 000 tonnes per year. Using the Fourier transform infrared spectroscopy technique, data were collected on 77 days during the period 1997–99. The relative number of samples that were above the detection limit was 15.2% and the mean value of the measurements was 10.5 ± 36.7 ppb [23.2 ± 81.1 µg/m³]. The maximum concentration measured was 3.1 ppm [6.8 mg/m³] (Chan et al., 2006). [Measurements were area samples and may underestimate exposure of the workers.]

In the monomer industry, potential exposure to compounds other than butadiene includes exposure to extraction solvents and components of the C₄ feedstock. Extraction solvents differ between facilities: common solvents include dimethylformamide, dimethylacetamide, acetonitrile, (β-methoxypropionitrile (Fajen, 1985a), furfural and aqueous cuprous ammonium acetate (Occupational Safety and Health Administration, 1990b). Stabilizers are commonly used to prevent the formation of peroxides in air and during polymerization. No information was available on these exposures or on exposure to chemicals other than butadiene that are produced in some facilities such as butylenes, ethylene, propylene, polyethylene and polypropylene resins, methyl-tert-butyl ether and aromatic hydrocarbons (Fajen, 1985b,c).

(b) Production of polymers and derivatives

In samples taken at a styrene–butadiene rubber plant in the USA in 1976 (Table 9), levels of butadiene above 100 ppm [220 mg/m³] were encountered by technical services personnel (115 ppm [253 mg/m³]) and an instrument man (174 ppm [385 mg/m³]; Meinhardt et al, 1978). At another styrene–butadiene rubber manufacturing plant in the USA in 1979, the only two departments in which levels were greater than 10 ppm [22 mg/m³] were the tank farm (53.4 ppm [118 mg/m³]) and maintenance (20.7 ppm [46 mg/m³]; Checkoway & Williams, 1982). Overall mean 8-h TWA exposure levels differed considerably between the two plants: 13.5 ppm [30 mg/m³] and 1.24 ppm [2.7mg/m³], respectively (Meinhardt et al., 1982).

Table 9

Eight-hour time-weighted average exposure levels of butadiene measured in two styrene–butadiene rubber manufacturing plants in the USA.

Detailed industrial hygiene surveys were conducted in 1986 in five of 17 facilities in the USA where butadiene was used to produce styrene–butadiene rubber, nitrile–butadiene rubber, polybutadiene rubber, neoprene and adiponitrile (Fajen, 1988). Levels of butadiene to which workers in various job categories were exposed are summarized in Table 10. Process technicians in unloading, in the tank farm and in the purification, polymerization and reaction areas, laboratory technicians and maintenance technicians were exposed to the highest levels. Short-term sampling showed that activities such as sampling a barge and laboratory work were associated with peak exposures of more than 100 ppm [220 mg/m³]. Full-shift area sampling indicated that geometric mean ambient concentrations of butadiene were less than 0.5 ppm [1.1 mg/m³] and usually less than 0.1 ppm [0.22 mg/m³] in all locations measured at the five plants.

Table 10

Eight-hour time-weighted average exposure levels in personal breathing zone samples at five plants that produced butadiene-based polymers and derivatives in the USA, 1986.

A biological monitoring study that used personal sampling reported average levels of butadiene of 0.30, 0.21 and 0.12 ppm [0.66, 0.46 and 0.27 mg/m³] for the high-, intermediate- and low-exposure groups, respectively, in a styrene–butadiene rubber plant in Texas, USA (Ward, J.B. et al., 1996).

In 13 of 27 European sites where styrene–butadiene rubber and styrene–butadiene latex were produced, less than 10% of the concentrations measured exceeded 5 ppm (Table 11; ECETOC, 1997).

Table 11

Eight-hour time-weighted average personal exposures to butadiene in styrene–butadiene rubber plants in the European Union (1984–93).

Data from the Netherlands are available from 1976 onwards, but the measurement methods used in the early surveys are unknown (Kwekkeboom, 1996; Dubbeld, 1998). No clear trend can be seen for the years 1990–97, but average exposures were relatively low (arithmetic mean < 3 ppm [6.6 mg/m³]) (Table 12).

Table 12

Eight-hour time-weighted average exposure levels of butadiene in personal breathing zone samples at a plant that produced styrene–butadiene polymer in the Netherlands, 1990–97.

Exposure of 38 workers was measured in a butadiene polymer production facility in China. Personal full-shift measurements established that workers in butadiene operations were exposed to a median level of 2.0 ppm [4.4 mg/m³]. Short-term breathing zone measurements of butadiene showed great extremes in exposure; DMF [dimethyl-formamide] analysts had a median exposure of 54 ppm [119 mg/m³] (range, below detection to 3090 ppm [6829 mg/m³]; 50 samples), polymer analysts had a median exposure of 6.5 ppm [14.4 mg/m³] (range, below detection to 1078 ppm [2382 mg/m³]; 41 samples) and maintenance-recovery workers had a median exposure of 7.0 ppm [15.5 mg/m³] (range, below detection to > 12 000 ppm [> 26 520 mg/m³]; 24 samples) (Hayes et al., 2001).

A biomonitoring study carried out in a styrene–butadiene rubber plant in Southeast Texas, in which 37 workers were monitored during their entire work shift using passive samplers, demonstrated that levels in the tank area exceeded the current Occupational Safety and Health Administation permissible exposure limit for butadiene. However, the workers wore protective equipment on this particular job. TWA values in various work areas are summarized in Table 13 (Ward et al., 2001).

Table 13

Time-weighted average exposures to butadiene in a styrene–butadiene rubber plant in the USA, 1998.

In 1998, 319 personal workshift TWA measurements of exposure to butadiene were obtained for 34 workers in a polymer production plant in the Czech Republic. The mean (± SD) concentration of butadiene was 1.8 ± 4.7 mg/m³ [0.8 ± 2.1 ppm]. The individual TWA measurements from all polymer production workers ranged from 0.002 to 39.0 mg/m³ [0.001–17.6 ppm]. The level of exposure of the control group was 0.03 ± 0.03 mg/m³ [0.01 ± 0.01 ppm], calculated from 28 personal TWA measurements (Albertini et al., 2003a).

A Finnish study assessed personal exposure to butadiene in three plants that manufactured styrene–butadiene latex. Full-shift air samples were collected from the breathing zone of 28 workers using passive samplers over 4 months. A total of 885 samples were collected and the number of samples per participant ranged from 19 to 39. Samples were collected at the same time in all three plants. The data showed that 624 (70.5%) of the samples were below the limit of quantification; 240 (27.1%) samples were between the limit of quantification and 1 ppm [2.2 mg/m³] and 21 (2.4%) were over the Finnish occupational exposure limit of 1 ppm [2.2 mg/m³]. Mean butadiene concentrations in the three plants were 0.068, 0.125 and 0.302 ppm [0.15, 0.28 and 0.67 mg/m³], respectively. Statistical analysis of the data did not indicate any significant difference between the plants when all results were considered (Anttinen-Klemetti et al., 2006).

In a Czech study that included 26 female control workers, 23 female butadiene-exposed workers, 25 male control workers and 30 male butadiene-exposed workers, 10 personal full-shift (8-h) measurements per worker over a 4-month period showed mean 8-h TWA exposure levels of 0.008 mg/m³ and 0.4 mg/m³ [0.004 and 0.18 ppm] for control and exposed women, respectively. The highest single 8-h TWA value among exposed women was 9.8 mg/m³ [4.5 ppm]. Mean 8-h TWA exposure levels were 0.007 mg/m³ and 0.8 mg/m³ [0.003 and 0.36 ppm] for control and exposed men, respectively; personal single 8-h TWA values of up to 12.6 mg/m³ [5.7 ppm] were measured in the exposed group. The concentrations for butadiene-exposed workers were significantly higher than those for the controls for both men and women; the concentrations for butadiene-exposed workers were significantly higher for men than for women (Albertini et al., 2007). [The difference in exposure levels may be due to differences in tasks performed by men and women.]

Data from a Canadian styrene–butadiene rubber plant indicate a clear decrease in exposure from 1977 to 1991 (Sathiakumar & Delzell, 2007; Table 14). The data were used to validate the estimates of historical exposure to butadiene (Macaluso et al., 1996, 2004; Sathiakumar et al., 2007).

Table 14

Exposure levels of butadiene in a styrene–butadiene rubber plant in Canada.

The manufacture of butadiene-based polymers and butadiene derivatives implies potential exposure to a number of other chemical agents that vary according to product and process and include other monomers (styrene, acrylonitrile, chloroprene), solvents, additives (e.g. activators, antioxidants, modifiers), catalysts, mineral oils, carbon black, chlorine, inorganic acids and caustic solutions (Fajen, 1986a,b; Roberts, 1986). Styrene, benzene and toluene levels were measured in 1979 in various departments of a plant that manufactured styrene–butadiene rubber in the USA: mean 8-h TWA levels of styrene were below 2 ppm [8.4 mg/m³], except for tank-farm workers (13.7 ppm [57.5 mg/m³], eight samples); mean benzene levels did not exceed 0.1 ppm [0.3 mg/m³], and those of toluene did not exceed 0.9 ppm [3.4 mg/m³] (Checkoway & Williams, 1982). Meinhardt et al. (1982) reported that the mean 8-h TWA levels of styrene in two styrene–butadiene rubber manufacturing plants were 0.94 ppm [3.9 mg/m³] (55 samples) and 1.99 ppm [8.4 mg/m³] (35 samples) in 1977; the average level of benzene measured in one of the plants was 0.1 ppm [0.3 mg/m³] (three samples). Average levels of styrene, toluene, benzene, vinyl cyclohexene and cyclooctadiene were reported to be below 1 ppm in another styrene–butadiene rubber plant in 1977 (Burroughs, 1977). Dimethyldithiocarbamate has been used in some plants and dermal exposure to this compound potentially exists (Delzell et al., 2001).

1.3.3. Environmental occurrence

According to the Environmental Protection Agency Toxic Chemical Release Inventory in the USA, industrial releases of butadiene to the atmosphere from industrial facilities in the USA were 4425 tonnes in 1987, 2360 tonnes in 1990 and 1385 tonnes in 1995. According to the same database, fugitive air emissions were 157 973 kg and point source air emissions were 450 926 kg in 2005 (Environmental Protection Agency Toxic Release Inventory, 2005; National Library of Medicine, 2008).

Under laboratory conditions, non-catalyst vehicles emitted butadiene at a rate of 20.7 ± 9.2 mg/kg. Vehicles that had a functioning catalyst–emission control device had an average emission rate of 2.1 ± 1.5 mg/km. Based on these numbers, the authors concluded that vehicle emissions of butadiene have been substantially underestimated (Ye et al., 1997). Based on an average of 20 000 km per year per car and approximately 243 million registered cars in the USA in 2004, and considering the average emission rates estimated by Ye et al. (1997), emissions of butadiene from automobile exhausts can be estimated to amount to approximately 106 770 tonnes per year.

Butadiene is also released to the atmosphere from the smoke of bush fires, the thermal breakdown or burning of plastics and by volatization from gasoline (Agency for Toxic Substances and Disease Registry, 1992; see IARC, 1992).

Kim et al. (2001, 2002) measured the concentrations of 15 volatile organic compounds, including butadiene, in a wide range of urban micro-environments in the United Kingdom (Table 15) and estimated the personal exposure of 12 urban dwellers directly and indirectly via static monitoring combined with a personal activities diary (Table 16).

Table 15

Mean concentrations of butadiene in microenvironments in the United Kingdom.

Table 16

Daytime and night-time concentrations (µg/m³) of butadiene recorded during personal exposure monitoring in the United Kingdom, 1999–2000.

Environmental exposure to emissions, including butadiene, was compared between bus and cycling commuters on a route in Dublin. Samples were collected during both morning and afternoon rush-hour periods using continuous sampling. The average concentrations experienced by the cyclist and the bus passenger for all journeys were 0.47 ppb [103 µg/m³] (SD, 0.19; min., 0.24; max., 0.81) and 0.78 ppb [1.7 µg/m³] (SD, 0.34; min., 0.34; max., 1.49), respectively (O'Donoghue et al., 2007).

In the United Kingdom, the estimated emission of butadiene in 1996 was 10.6 thousand tonnes. Road vehicle exhaust emissions dominated and comprised 68% of the total emissions, while emissions from off-road vehicles and machinery accounted for 14%. The remaining emissions arose from the chemical industry, during the manufacture of butadiene and its use in the production of various rubber compounds. These two processes accounted for 8 and 10%, respectively, of total emissions in the United Kingdom in 1996 (Dollard et al., 2001).

Municipal structural fires are a source of butadiene, and the mean level of butadiene from nine fires ranged from 0.03 to 4.84 ppm [0.07–9.9 mg/m³] (Austin et al., 2001). Domestic wood burning also has an impact on levels of butadiene in homes. Wood burners had a significantly higher personal exposure to butadiene (median, 0.18 µg/m³) than the reference group. Similarly, significantly higher indoor levels were reported (median, 0.23 µg/m³) in homes of wood burners than in the homes of the reference group (Gustafson et al., 2007).

The intake of butadiene that results from exposure to environmental tobacco smoke for a person who lives with one or more smokers in homes where smoking is permitted was estimated to be in the range of 16–37 µg per day (Nazaroff & Singer, 2004). The levels of butadiene in public houses in Dublin were assessed before and after the smoking ban in 2004. The average level before the ban was 4.15 µg/m³ [1.87 ppb]. The levels of butadiene recorded in the same establishments when cigarettes were no longer being smoked dropped significantly to 0.22 µg/m³ [0.1 ppb], which is still higher than the average ambient level (0.12 µg/m³ [0.05 ppb]) (McNabola et al., 2006).

In the metropolitan area of Mexico City, three persons who were simultaneously monitored for butadiene inside the home and outdoors had median levels of 2.1 µg/m³ [1 ppb] (max., 11.5 µg/m³ [5.2 ppb]), 2.0 µg/m³ [0.9 ppb] (max., 8.3 µg/m³ [3.7 ppb]) and 0.8 µg/m³ [0.4 ppb] (max., 4.6 µg/m³ [2.1 ppb]) for personal, indoor and outdoor exposure, respectively (Serrano-Trespalacios et al., 2004).

Ambient concentrations of butadiene were measured in Japan during the years 1997–2003 at general environmental stations, roadside stations and industrial vicinity stations. The mean levels in 1998 were 0.28, 0.56 and 0.37 µg/m³ [0.13, 0.25 and 0.17 ppb] for the general environment, roadside and industrial vicinity, respectively. The overall level was 0.36 µg/m³ [0.16 ppb]. In 2003, corresponding levels were 0.22, 0.42 and 0.31 µg/m³ [0.10, 0.19 and 0.14 ppb], with an overall level of 0.29 µg/m³ [0.13 ppb] (Higashino et al., 2006).

Mainstream and sidestream cigarette smoke contain approximately 20–40 mg and 80–130 µg butadiene per cigarette, respectively; levels of butadiene in smoky indoor environments are typically 10–20 µg/m³ [5–9 ppb] (IARC, 2004).

Based on its physical and chemical properties, butadiene is unlikely to be detected in water or in soil (Agency for Toxic Substances and Disease Registry, 1992).

2.2.1. Monomer production

A cohort mortality study included men who were assigned to any of three butadiene production units located within several chemical plants in the Kanawha Valley of West Virginia, USA. Of the 364 men included in the study, 277 (76%) were employed in a ‘Rubber Reserve’ plant that operated during the Second World War (Ward et al., 1995). The plants produced butadiene from ethanol or from olefin cracking. The butadiene production units included in this study were selected from an index of chemical departments that was developed by the Union Carbide Corporation and included only departments where butadiene was a primary product and neither benzene nor ethylene oxide was present. The cohort studied was part of a large cohort of 29 139 chemical workers whose mortality experience had been reported earlier, although without regard to specific exposures (Rinsky et al., 1988). Three subjects were lost to follow-up (0.8%). Mortality from all cancers was not increased (48 deaths; standardized mortality ratio [SMR], 1.1; 95% confidence interval [CI], 0. 8–1.4). Seven deaths from lymphatic and haematopoietic cancers occurred (SMR, 1.8; 95% CI, 0.7–3.6), including four cases of lymphosarcoma and reticulosarcoma (SMR, 5.8; 95% CI, 1.6–14.8). Three cases had a duration of employment of 2 years or more (SMR, 8.3; p < 0.05). Two cases of leukaemia (SMR, 1.2; 95% CI, 0.2–4.4) also occurred. A non-significant excess of mortality from stomach cancer was observed (SMR, 2.4; 95% CI, 0.8–5.7). All five cases of stomach cancer occurred among the subset of workers who had been employed in the ‘Rubber Reserve’ plant for 2 years or more (SMR, 6.6; 95% CI, 2.1–15.3).

The mortality of a cohort of workers who manufactured butadiene monomer in Texas, USA (Downs et al., 1987), has been investigated repeatedly with updated and extended follow-up (Divine, 1990; Divine et al., 1993; Divine & Hartman, 1996). The latest available update, that included 5 additional years of follow-up up to 31 December 1999, was reported by Divine and Hartman (2001). The cohort at that time included 2800 male workers (of whom 216 were non-white and 10 were of unknown race) who had been employed for at least 6 months between 1943 and 1996. Exposure assessment was based on job history and industrial hygiene sampling data for the years after 1981. Each job was assigned a score for exposure to butadiene that took into account calendar period and type of operation. No information was reported on exposure to chemicals other than butadiene. The number of workers lost to follow-up was 192 (6.7%), all but 17 (< 1%) of whom were known to be alive at the end of 1998. A total of 1422 deaths were identified through to 1999, and death certificates were obtained for all but 19 (1.3%) of the deaths. SMRs were calculated using mortality rates for the US population as the reference. The SMR for all causes of death was 0.9 (95% CI, 0.8–0.9) and that for all cancers (333 deaths) was 0.9 (95% CI, 0.8–1.0). Fifty deaths from lymphatic and haematopoietic cancers (International Classification of Diseases [ICD]-8, 200–209; SMR, 1.4; 95% CI, 1.1–1.9), nine deaths from lymphosarcoma and reticulosarcoma (ICD-8, 200; SMR, 2.0; 95% CI, 0.9–3.9), 19 deaths from non-Hodgkin lymphoma (ICD-8, 200, 202; SMR, 1.5; 95% CI, 0.9–2.3), four deaths from Hodgkin lymphoma (ICD-8, 201; SMR, 1.6; 95% CI, 0.4–4.1), 18 deaths from leukaemia (ICD-8, 204–207; SMR, 1.3; 95% CI, 0.8–2.0), seven deaths from multiple myeloma (ICD-8, 203; SMR, 1.3; 95% CI, 0.5–2.6) and 18 deaths from cancer of other lymphatic tissue (ICD-8, 202, 203, 208; SMR, 1.3; 95% CI, 0.8–2.1) were observed. However, the latter category overlapped with non-Hodgkin lymphoma and multiple myeloma. The SMRs for the lymphatic and haematopoietic cancers did not increase with length of employment. Analysis by date of employment showed an increased risk for lymphatic and haematopoietic cancers among those first employed between 1942 and 1949. A separate mortality analysis for non-whites showed lower than expected mortality for all malignant neoplasms (17 observed, 19 expected) and a single death from lymphatic and haematopoietic cancer. Subcohort analyses were made for groups that were classified as having background, low and varied exposure. The background-exposure group in-cluded persons in offices, transportation, utilities and warehouses; the low-exposure group had spent some time in operating units; and the varied-exposure group included those with greatest potential exposure in operating units, laboratories and maintenance. In the background-exposure group, four deaths from lymphatic and haematopoietic cancers (ICD-8, 200–209) were observed among those employed for < 5 years (SMR, 1.9; 95% CI, 0.5–4.7) and four among those exposed for > 5 years (SMR, 1.7; 95% CI, 0.5–4.3). Eleven deaths from lymphatic and haematopoietic cancers (ICD-8, 200–209) were observed in the low-exposure group, seven of which were among those with < 5 years of employment (SMR, 0.9; 95% CI, 0.4–1.9) and four among those employed for > 5 years (SMR, 0.6; 95% CI, 0.2–1.6). In the varied-exposure group, with the highest potential for exposure to butadiene, 34 deaths from lymphatic and haematopoietic cancers (ICD-8, 200–209) were observed, 20 of which were among those employed for < 5 years (SMR, 1.8; 95% CI, 1.1–2.8) and 14 among those exposed for > 5 years (SMR, 1.5; 95% CI, 0.8–2.5). In all groups, the SMRs for lymphatic and haematopoietic cancer decreased with duration of employment. For lymphosarcoma and reticulosarcoma, two deaths occurred in the low-exposure group (one among those employed < 5 years and one among those employed > 5 years) and seven deaths were observed in the varied-exposure group, five of which were among those employed for < 5 years (SMR, 3.7; 95% CI, 1.2–8.7) and two among those employed for > 5 years (SMR, 1.87; 95% CI, 0.23–6.76). Three deaths from leukaemia occurred (SMR, 0.7; 95% CI, 0.1–2.0) in the low-exposure subgroup and 13 cases were observed in the varied-exposure group, eight of which were among those employed for < 5 years (SMR, 1.9; 95% CI, 0.8–3.7) and five among those employed > 5 years (SMR, 1.4; 95% CI, 0.4–3.2). Six deaths from non-Hodgkin lymphoma were observed in the low-exposure group, four among short-term employees (SMR, 1.5; 95% CI, 0.4–3.8) and two among those employed for > 5 years (SMR, 0.9; 95% CI, 0.1–3.2), and 12 deaths occurred in the varied-exposure group, eight of which were among those employed for < 5 years (SMR, 2.0; 95% CI, 0.9–3.9) and four among long-term employees (SMR, 1.1; 95% CI, 0.3–2.9). The ‘varied-exposure’ group with high potential for exposure to butadiene showed elevated SMR estimates for all subcategories of lymphatic and haematopoietic cancers, but the increase was statistically significant only for lympho/reticulosarcoma among those employed for < 5 years. Slightly elevated SMRs were also found in the low-exposure group for cancer of the kidney (three cases; SMR, 1.6; 95% CI, 0.3–4.6; and three cases; SMR, 1.9; 95% CI, 0.4–5.4; among short- and long-term employees, respectively). In the varied-exposure group, a suggestive excess incidence of kidney cancer was only present among those employed for > 5 years (four cases; SMR, 1.65; 95% CI, 0.45–4.22). Survival analysis by Cox regression was carried out using a cumulative exposure score as a time-dependent explanatory variable for all lymphohaematopoietic neoplasms, non-Hodgkin lymphoma and leukaemia. None of these cancers was significantly associated with the cumulative exposure score. The elevated risk for all the lymphohaematopoietic cancers and their subcategories occurred among persons who were first employed before 1950. [The Working Group noted that although there was no evidence of an exposure–response relationship, it is probable that many workers during the years of the Second World War would have had short but relatively intense exposures, and thus duration of exposure may not be the most relevant dose metric.]

Another relatively small retrospective mortality study, together with a prospective morbidity survey, was performed on male employees at the Shell Deer Park Manufacturing Complex in the USA (Cowles et al., 1994) and was updated with a 9-year extension of the follow-up (Tsai et al., 2001). Butadiene monomer production took place in the facility between 1941 and 1948 and from 1970 onwards. The cohort comprised 614 eligible male employees who had worked for 5 years or more in jobs that entailed potential exposure to butadiene. Also eligible were employees who had worked for at least half of their total duration of employment during 1948–89 in a job that entailed potential exposure to butadiene (with a minimum 3-month period in such jobs). Female employees were excluded because of the small number (35) who met the eligibility criteria. Industrial hygiene data from 1979 to 1992 showed that few exposures to butadiene exceeded 10 ppm [22 mg/m³] as an 8-h TWA and that most were below 1 ppm [2.2 mg/m³]; the arithmetic mean exposure was 3.5 ppm [7.7 mg/m³]. Only one study member had unknown vital status at the end of the follow-up. Person–years were accrued after 1 April 1948 from the time that a person first met the eligibility criteria. Death certificates were obtained for all known decedents. SMRs adjusted for age, race and calendar year were calculated using county-specific mortality rates as the reference. Six hundred and fourteen cohort members contributed a total of 12 391 person–years during the expanded study period, during which 61 deaths were identified. The SMR for all causes of death was 0.55 (95% CI, 0.42–0.70) and that for all malignant neoplasms was 0.6 (16 deaths; 95% CI, 0.3–0.9). Eight deaths were due to lung cancer (SMR, 0.7; 95% CI, 0.2–3.1) and three to cancer of the lymphatic and haematopoietic tissues (SMR, 1.1; 95% CI, 0.3–1.5). No deaths from leukaemia were observed, whereas one death was expected. The morbidity study included 289 of the 614 cohort members who were actively employed at some time between 1 January 1992 and 31 December 1998. The morbidity experience of this group was compared with that of an internal comparison group of 1386 active employees during the same period who had had no exposure to butadiene. A morbidity event was defined as an absence from work of > 5 days during 1992–98 that resulted from a specific diagnosed disorder. No meaningful differences in morbidity events between the butadiene-exposed and unexposed employees in the rest of the Shell Deer Park Manufacturing Complex were observed. [The Working Group noted that one criteria for inclusion in the cohort (at least half of total employment during 1948–89 in a potentially exposed job) was a potential source of bias, and that the SMR for all causes was unusually low.]

2.2.2. Styrene–butadiene rubber production

The 9-year mortality experience of a cohort of 6678 male rubber workers from a single, large tyre manufacturing plant in Ohio, USA, approximately 4% of whom worked in the manufacture of synthetic rubber, was investigated during 1964–72 (McMichael et al., 1974, 1976). Death rates from various specific causes were increased and included lymphatic and haematopoietic cancers in general (43 observed deaths; SMR, 1.36), lymphosarcoma and Hodgkin lymphoma (15 observed; SMR, 1.64) and leukaemia (17 observed; SMR, 1.26). A case–cohort study was nested within the cohort to investigate the association of excesses of mortality with specific jobs within the rubber industry to compare workers who died from cancers in the 10-year period 1964–73 with a sample of members of the whole cohort and to elucidate differences in work histories (McMichael et al., 1976). A 6.2-fold increase in risk for lymphatic and haematopoietic cancers (99.9% CI, 4.1–12.5) and a 3.9-fold increase in risk for lymphatic leukaemia (99.9% CI, 2.6–8.0) were found in association with more than 5 years of work in manufacturing units that produced mainly styrene–butadiene rubber during 1940–60. [The Working Group noted that no information was provided on exposure to specific substances including potentially confounding chemicals such as benzene.]

Meinhardt et al. (1982) studied the mortality experience of white male workers who had been employed for at least 6 months in a two-plant complex styrene–butadiene rubber facility in the USA. A total of 1662 workers employed in Plant A between 1943 and 1976 and 1094 workers employed in Plant B between 1950 and 1976 were followed up through to 31 March 1976. Nine deaths from cancer of the lymphatic and haematopoietic tissues (ICD-7, 200–205) were seen in workers in Plant A (SMR, 1.6). The SMR among those from Plant A who were first employed between January 1943 and December 1945 was 2.1. Five deaths from leukaemia (ICD-7, 204) were observed in Plant A among workers employed between 1943 and 1945 (SMR, 2.8). In Plant B, two deaths from lymphatic and haematopoietic neoplasms (one lymphosarcoma/reticulosarcoma and one leukaemia) were observed.

Matanoski et al. (1990) investigated mortality patterns from 1943 (synthetic rubber production began in 1942) through to 1982 among 12 113 employees at styrene–butadiene rubber plants in Canada and the USA who had previously been followed up through to 1979 by Matanoski and Schwartz (1987). Overall, there were no increases in mortality from lymphatic and haematopoietic cancers. When employees were classified according to their longest-held job, production workers (presumed by the authors to be those with highest exposures to butadiene) had a significant excess of ‘other lymphatic cancer’ (nine deaths; SMR, 2.60; 95% CI, 1.19–4.94). When mortality among production workers was examined by race, a significant excess for leukaemia was seen in blacks (three deaths; SMR, 6.56; 95% CI, 1.35–19.06). Of 92 deaths among black production workers, six were from all lymphohaemopoietic cancers (SMR, 5.07; 95% CI, 1.87–11.07).

Nested case–control studies were conducted within the styrene–butadiene rubber cohort in the USA and Canada (Santos-Burgoa et al., 1992; Matanoski et al., 1997). In the study by Santos-Burgoa et al. (1992), 59 cases of lymphatic and haematopoietic cancer in male workers (1943–82) were matched to 193 controls by plant, age, year of hire, duration of employment and survival to time of death of the case. Each job was assigned an estimated rank of exposure to butadiene and styrene, and cumulated exposure for each worker was calculated using employment histories. A strong association was identified for both butadiene (odds ratio, 9.4; 95% CI, 2.1–22.9) and styrene (odds ratio, 3.1; 95% CI, 0.8–11.2). After controlling for the other exposure, the odds ratio for exposure to butadiene remained high and significant (odds ratio, 7.4; 95% CI, 1.3–41.3) whereas the relative risk estimate for styrene was approximately unity (odds ratio, 1.1; 95% CI, 0.23–4.95).

Matanoski et al. (1997) conducted a second case–control study that was nested in the styrene–butadiene cohort and included as cases most of the same lymphatic and haematopoietic cancer decedents studied by Santos-Burgoa et al. (1992). In this study, hospital records obtained for 55 of the 59 cases studied by Santos-Burgoa et al. (1992) were reviewed to confirm death certificate reports of lymphatic and haematopoietic cancer. The review confirmed all leukaemias, eliminated two cases and added one case of non-Hodgkin lymphoma and confirmed all cases of Hodgkin lymphoma and multiple myeloma. The final case groups included 58 total lymphatic and haematopoietic cancers, 12 non-Hodgkin lymphomas (seven lymphosarcomas and five other non-Hodgkin lymphomas), eight Hodgkin lymphomas, 26 leukaemias and 10 multiple myelomas. Controls were 1242 employees who were chosen to reflect the distribution of the cohort by plant and age, who had to have lived at least as long as cases and who represented approximately 1% of the cohort, but were not matched individually to cases. Quantitative exposure estimates for butadiene and styrene were developed by using exposure measurements for work areas and jobs, when available, and a modelling procedure to obtain estimates for jobs that had no measurements. Plant- and work area-specific exposure estimates were linked to work histories to obtain indices of cumulative exposure (ppm–months) and average intensity of exposure (ppm). Odds ratios for an average intensity of exposure of 1 ppm compared with 0 ppm and for ppm–months as a continuous variable were estimated using logarithmically transformed exposure data in unconditional logistic regression models that controlled for year of birth, period of hire, age at hire, race and length of employment. Leukaemia and Hodgkin lymphoma were associated positively with average intensity of exposure to butadiene (odds ratio at 1 ppm: leukaemia, 1.50; 95% CI, 1.07–2.10; Hodgkin lymphoma, 1.7; 95% CI, 0.99–3.02) and with ppm–months of exposure to butadiene. Non-Hodgkin lymphoma and multiple myeloma were associated positively with average intensity of exposure and cumulative exposure to styrene but not with indices of exposure to butadiene. Further analyses indicated that, in models that included both average intensity of exposure to butadiene and an indicator for longest employment in service, labour and laboratory work areas, both variables were statistically significantly, positively associated with leukaemia. Separate analyses of lymphoid leukaemia (10 cases) and myeloid leukaemia (15 cases) found that the average intensity of exposure to butadiene, but not work area, was significantly associated with lymphoid leukaemia. Average intensity of exposure to butadiene and work area were both associated positively with myeloid leukaemia, but the association was significant only for work area. Matanoski et al. (1997) suggested that misclassi-fication of quantitative indices of exposure to butadiene could explain the latter results.

Delzell et al. (1996) and Sathiakumar et al. (1998) evaluated the mortality experience of 15 649 men (87% white and 13% black) who had been employed for at least 1 year at any of eight styrene–butadiene rubber plants in the USA and Canada and who had worked in styrene–butadiene rubber-related operations in these plants. Seven of the plants had been studied previously by Matanoski and Schwartz (1987), Matanoski et al. (1990), Santos-Burgoa et al. (1992) and Matanoski et al.(1993, 1997); the two-plant complex studied earlier by Meinhardt et al. (1982) and Lemen et al. (1990) was also included. Complete work histories were available for 97% of the subjects: about 75% was exposed to butadiene and 83% was exposed to styrene. A list was developed to identify every combination of work area and job title, for a total of 8281 unique combinations. Using information from the plant on processes and operations and on jobs and tasks within each type of operation, 308 groups of work area were specified, and comprised processes and jobs that were considered to be similar. For analysis, these were further classified into five main process groups and seven process subgroups. During 1943–91, the cohort had a total of 386 172 person–years of follow-up and 734 individuals were lost to follow-up (5%). A total of 3976 deaths were observed compared with 4553 deaths expected on the basis of general population mortality rates for the USA or (for the Canadian subcohort) Ontario (SMR, 0.87; 95% CI, 0.85–0.90). Mortality from cancer was slightly lower than expected (950 deaths; SMR, 0.93; 95% CI, 0.87–0.99). Eleven lymphosarcomas were observed (SMR, 0.80; 95% CI, 0.40–1.40) and 42 ‘other lymphopoietic cancers’ (SMR, 0.97; 95% CI, 0.70–1.52) which included 17 non-Hodgkin lymphomas, eight Hodgkin lymphomas, 14 multiple myelomas, one polycythaemia vera and two myelofibroses. In addition, 48 deaths from leukaemia were observed in the full cohort (SMR, 1.31; 95% CI, 0.97–1.74), including 45 among ‘ever hourly paid’ (86% of the cohort) subjects (SMR, 1.43; 95% CI, 1.04–1.91). The excess was fairly consistent across plants and was concentrated among ‘ever hourly paid’ subjects with 10 or more years of employment and 20 or more years since hire (28 deaths; SMR, 2.24; 95% CI, 1.49–3.23) and among polymerization workers (15 deaths; SMR, 2.51; 95% CI, 1.40–4.14), maintenance labourers (13 deaths; SMR, 2.65; 95% CI, 1.41–4.53) and laboratory workers (10 deaths; SMR, 4.31; 95% CI, 2.07–7.93). Polymerization workers and maintenance labourers had potentially high exposure to butadiene but only low-to-moderate exposure to styrene. Sathiakumar et al. (1998) reported that, among subjects with ≥ 10 years of employment and ≥ 20 years since hire, moderately non-significantly increased mortality from non-Hodgkin lymphoma was also apparent (15 deaths; SMR, 1.37; 95% CI, 0.77–2.26), but without any consistent pattern by duration of employment, time since hire, period of hire or process group. Mortality from other types/sites of cancer was not significantly elevated in this cohort.

Macaluso et al. (1996) reported an additional analysis, with more detailed exposure assessment, of mortality from leukaemia among 16 610 subjects (12 412 exposed to butadiene) employed at six of the eight North American styrene–butadiene rubber manufacturing plants investigated by Delzell et al. (1996) [14 295 workers were included but a further 2350 workers from the same plants who were not employed in styrene–butadiene rubber operations at those plants were not included in Delzell et al. (1996)]. A total of 418 846 person–years of follow-up through 1991 and 58 leukaemia deaths, seven of which were reported to be the contributory (‘underlying’) cause of death, were included only in analyses that used internal comparisons. Retrospective quantitative estimates of exposure to butadiene, styrene and benzene were developed and the estimation procedure identified work areas within each manufacturing process, historical changes in exposure potential and specific tasks that involved exposure. Mathematical models were then used to calculate job- and time period-specific average exposures. The resulting estimates were linked with work histories to obtain cumulative exposure estimates that were employed in stratified and Poisson regression analyses of mortality rates. Mantel-Haenszel rate ratios adjusted by race, age and cumulative exposure to styrene increased with cumulative exposure to butadiene from 1.0 in the unexposed category through to 2.0, 2.1, 2.4 and 4.5 in the estimated exposure categories 0, < 1, 1–19, 20–79 and ≥ 80 ppm–years, respectively (p for trend = 0.01). The trend of increasing risk with exposure to butadiene was still significant after exclusion of the unexposed category (p = 0.03). The risk pattern was less clear and statistically non-significant for exposure to styrene (rate ratios, 0.9, 5.4, 3.4 and 2.7 in the estimated exposure categories of < 5, 5–9, 10–39 and ≥ 40 ppm–years, respectively; p for trend = 0.14) and the association with benzene disappeared after controlling for exposure to butadiene and styrene.

Irons and Pyatt (1998) suggested that apparently different patterns of risk for leukaemia between workers employed in butadiene monomer production and those involved in styrene–butadiene rubber production might be linked to a class of chemicals with haematotoxic and immunotoxic potential (dithiocarbamates) that were present in styrene–butadiene rubber but not butadiene monomer production. In particular, they suggested that dimethyldithiocarbamate, which was used between the early 1950s and 1965 in the majority of styrene–butadiene rubber plants as a stopping agent in the cold polymerization reaction, might confound an association between butadiene and leukaemia in exposed workers.

A further analysis was then conducted among the North American synthetic rubber industry workers to evaluate the relative relevance of butadiene, styrene and dimethyl-dithiocarbamate in the statistically significantly increased risk for mortality from leukaemia (Delzell et al., 2001). The analysis included 13 130 men who had been employed for at least 1 year during 1943–91 at any of six synthetic rubber plants (of the eight previously studied by Delzell et al. (1996) and Sathiakumar et al. (1998)) that had sufficient information for quantitative exposure estimation. Revised exposure estimates for butadiene and styrene and new quantitative estimates for dimethyldithiocarbamate were developed (Macaluso et al., 2004). Quantitative estimates of cumulative exposure were obtained by (a) identifying work area/job groups (‘jobs’) that consisted of homogeneous work activities; (b) identifying the tasks that comprised each job; (c) identifying historical changes in exposure potential for each task; (d) computing time period-specific average exposure concentrations in parts per million for tasks and jobs; (e) compiling the job-specific estimates into a job–exposure matrix for each plant; and (f) linking the resulting job–exposure matrices with work histories to obtain cumulative exposure estimates. The job–exposure matrix contained plant-specific estimates of 8-h TWA exposure concentrations for butadiene and other chemicals for each job and for each year from 1943 through to 1991. Mathematical models were used to calculate plant-, task-, work area/job group- and time period-specific average exposures. The revised TWA exposure estimates, compared with the original estimates, were about four to six times higher for butadiene and two times higher for styrene. Estimates of the annual number of peaks of exposure in each job (i.e. for each job, the number of component tasks in which intensity exceeded 100 ppm for butadiene and 50 ppm for styrene) were also computed. Vital status at the end of 1991 was known for over 99% of subjects. A total of 234 416 person–years of observation (average, 18 person–years per subject) were accrued during the follow-up period and included in the analysis. Information from death certificates was available for 3813 (98%) of 3892 decedents; the death certificates of 58 decedents mentioned leukaemia as the underlying or contributing cause of death. For 48 of these, medical records confirmed that they had had leukaemia; for 10, medical records were not available and these were retained in the case group. An additional decedent had medical records that indicated leukaemia but the death certificate mentioned myelodysplasia: he was added to the case group (total, 59 decedents). Poisson regression analysis was used to estimate relative rates of mortality from leukaemia and their 95% CIs for a particular agent/exposure category compared with the category of workers who were unexposed or had low exposure; regression models for each agent took into account the level of exposure, age, years since hire and, in some models, co-exposure to one or both of the other chemicals. In some analyses, exposure was lagged under the assumption that exposures that occurred within 5 or 10 years before death were etiologically irrelevant. Mortality from leukaemia showed a consistently positive association with increasing exposure to butadiene (relative rate, 1.0; 1.2; 95% CI, 0.5–3.0; 2.0; 95% CI 0.8–4.8; and 3.8; 95% CI, 1.6–9.1; for exposure to 0, 0–86.3, 86.3–362.2 and ≥ 362.2 ppm–years, respectively) [p for trend < 0.001] and to styrene (relative rate, 1.0; 1.2; 95% CI, 0.5–3.3; 2.3; 95% CI, 0.9–6.2; and 3.2; 95% CI, 1.2–8.8; for exposure to 0, 0–20.6, 20.6–60.4 and ≥ 60.4 ppm–years, respectively) [p for trend = 0.001]. Exposure to dimethyldithiocarbamate also showed consistently increased relative rates, but with no monotonic trend. After controlling for exposure to styrene and dimethyldithiocarbamate as well as for age and years since hire, exposure to butadiene remained consistently but not statistically significantly associated with increasing mortality from leukaemia (relative rate, 1.0; 1.3; 95% CI, 0.4–4.3; 1.3; 95% CI, 0.4–4.6; and 2.3; 95% CI, 0.6–8.3) [p for trend = 0.250].

When exposure to total peaks (> 100 ppm for butadiene and > 50 ppm for styrene) was used in the analysis, a positive association of exposure to butadiene and styrene with mortality from leukaemia was again apparent; adjustments for other agents made the associations irregular and imprecise. Lagging for a 5- or 10-year exposure period did not alter the results materially. Models were also run to estimate relative rates for leukaemia for exposures to butadiene calculated at intensities of < 100 ppm and > 100 ppm. In the former analysis, a statistically non-significant trend of increasing relative rate with increasing number of butadiene ppm–years was found (relative rate, 1.0; 1.1; 95% CI, 0.5–2.7; 2.8; 95% CI, 1.2–6.8; and 3.0; 95% CI, 1.2–7.1; for exposure to 0, 0–37.8, 37.8–96.3 and ≥ 96.3 ppm–years, respectively; p for trend = 0.25), whereas the trend for exposure to > 100 ppm was statistically significant (relative rate, 1.0; 2.1; 95% CI, 0.9–5.1; 2.8; 95% CI, 1.2–6.7; and 5.8; 95% CI, 2.4–13.8; for exposure to 0, 0–46.5, 46.5–234.3 and ≥ 234.3 ppm–years, respectively; p for trend = 0.01). Analyses based on quartiles or quintiles of exposure yielded patterns of results similar to those seen in the analyses based on tertiles shown above. Exposure to butadiene and dimethyldithiocarbamate was further categorized to study interaction, but no clear interaction between the two agents was apparent. The analysis confirmed a significant, although weaker, association with exposure to butadiene even after controlling for dimethyldithiocarbamate: the relative rate for leukaemia for exposure to 0–38.7, 38.7–287.3 and ≥ 287.3 ppm–years of butadiene was, respectively, 1.0, 1.5 (95% CI, 0.8–2.9) and 3.4 (95% CI, 1.8–6.4) when unadjusted for exposure to dimethyldithiocarbamate (p for trend = 0.0001) and 1.0, 1.0 (95% CI, 0.5–2.1) and 2.0 (95% CI, 0.9–4.3) when adjusted for exposure to dimethyldithiocarbamate (p for trend = 0.007). Dimethyldithiocarbamate was associated with a non-monotonically increasing risk and the trend was statistically nonsignificant after controlling for butadiene. The relative rate for exposure to 0–342.4, 342.4–1222.6 and ≥ 1222.6 mg–years/cm of dimethyldithiocarbamate was, respectively, 1.0, 3.6 (95% CI, 1.9–6.7) and 2.9 (95% CI, 1.5–5.3) when unadjusted for exposure to butadiene (p for trend = 0.003) and 1.0, 3.2 (95% CI, 1.6–6.3) and 2.1 (95% CI, 1.0–4.4) when adjusted for exposure to butadiene (p for trend = 0.196). A similar analysis was conducted for co-exposure to butadiene and styrene. The marginal trend was consistent and statistically significant for exposure to butadiene adjusted for styrene (p for trend = 0.006), but not for styrene adjusted for butadiene (p for trend = 0.763). Analyses by specific subgroups of leukaemia were conducted but sparse data rendered the results largely uninformative. [The Working Group noted the difficulties in estimating exposures to dimethyldithiocarbamate, which are primarily dermal, and that substantial misclassification of exposure to this chemical was possible. Furthermore, the assessment of exposure to dimethyldithiocarbamate was performed with the knowledge of which departments had excess mortality from leukaemia and could conceivably have been biased by this knowledge.]

Mortality from lymphatic and haematopoietic cancer in the updated North American cohort was studied in relation to exposure to butadiene, styrene and dimethyldithiocarbamate (Graff et al., 2005). Two of the US plants were not included in this analysis because information on work area/job group was not sufficient for quantitative exposure estimation for all substances. Included were 16 579 men who had worked at any of the six study plants for at least 1 year by the end of 1991 and who were followed up between 1944 and 1998. All work histories and exposure data came from the previous study of Delzell et al. (2001), and exposure estimation procedures were those described by Macaluso et al. (2004). Information on vital status through to 1998 was established for 97% of the study group. Cause of death was ascertained through death certificates, the national death index and a search of medical records. Most analyses used Poisson regression models to obtain maximum likelihood estimates of the relative rate for the contrast between categories of one agent, adjusting for other agents and for additional potential confounders. SMR analyses by level of exposure to the agent were also made using state-specific US and Canadian male mortality rates as the comparison group and data from death certificates and the national death index only to determine causes of death. During the observation period, 500 174 person–years were accrued. Based on a review of medical records where possible, 81 deaths from leukaemia, 58 from non-Hodgkin lymphoma, 27 from multiple myeloma and 13 from Hodgkin disease were ascertained. Single-agent analyses, adjusting for age and years since hire, indicated a positive association between exposure to butadiene and leukaemia (relative risk, 1.0; 1.4; 95% CI, 0.7–3.1; 1.2; 95% CI, 0.6–2.7; 2.9; 95% CI, 1.4–6.4; and 3.7; 95% CI, 1.7–8.0; for exposure to 0, 0–33.7, 33.7–184.7, 184.7–425.0 and ≥ 425.0 ppm–years, respectively) [p for trend < 0.001] and between exposure to styrene and leukaemia (relative risk, 1.0, 1.3; 95% CI, 0.6–3.2; 1.6; 95% CI, 0.7–3.9; 3.0; 95% CI, 1.2–7.1; and 2.7; 95% CI, 1.1–6.4; for exposure to 0, 0–8.3, 8.3–31.8, 31.8–61.1 and ≥ 61.1 ppm–years, respectively) [p for trend = 0.001]. Exposure to dimethyldithiocarbamate was also positively associated, but with no monotonic increase (relative risk, 1.0; 2.5; 95% CI, 1.2–5.0; 3.0; 95% CI, 1.5–5.9; 4.9; 95% CI, 2.5–9.7; and 2.7; 95% CI, 1.4–5.4; for exposure to 0, 0–185.3, 185.3–739.4, 739.4–1610.3 and ≥ 1610.3 mg–years/cm, respectively) [p for trend = 0.001]. Similar results were obtained for butadiene and styrene when total peaks were used as exposure metrics. Exposures to butadiene, styrene and dimethyldithiocarbamate were found to be highly correlated. In the models that also controlled for other agents, the estimated relative rates for the increasing categories of butadiene ppm–years were 1.0, 1.4 (95% CI, 0.5–3.9), 0.9 (95% CI, 0.3–2.6), 2.1 (95% CI, 0.7–6.2) and 3.0 (95% CI, 1.0–9.2) [p for trend = 0.028]; for styrene, the association appeared to be negative [p for trend = 0.639]; and for dimethyldithiocarbamate, a positive association was still apparent with no monotonic exposure–response pattern (relative risk, 1.0, 2.6; 95% CI, 1.2–5.8; 3.1; 95% CI, 1.4–7.1; 4.4; 95% CI, 1.9–10.2; and 2.0; 95% CI, 0.8–4.8) [p for trend = 0.066]. Individual adjustment for one or both of the other main exposure factors yielded similar results: the positive increasing association between exposure to butadiene and leukaemia was slightly reduced; no association remained for styrene; dimethyldithiocarbamate still exhibited a positive association with leukaemia but with no dose–response. Leukaemia was positively associated with butadiene ppm–years accrued at exposure intensities > 100 ppm and < 100 ppm. There was no evidence of an interaction between butadiene and dimethyldithiocarbamate or between butadiene and styrene. Analyses by histological type clearly showed an association of butadiene ppm–years with chronic lymphocytic leukaemia (relative risk, 1.0; 1.5; 95% CI, 0.6–4.0; 3.9; 95% CI, 1.3–11.0) [p for trend = 0.014], chronic myelogenous leukaemia (relative risk, 1.0; 2.7; 95% CI, 0.7–10.4; 7.2; 95% CI, 1.7–30.5) [p for trend = 0.007] and other leukaemia (relative risk, 1.0; 1.1; 95% CI, 0.3–3.9; 4.0; 95% CI, 1.0–15.0) [p for trend = 0.060] for exposures to 0–33.7, 33.7–425.0 and ≥ 425.0 ppm–years, respectively. No associations were found for non-Hodgkin lymphoma, multiple myeloma or acute myelogenous leukaemia. Results for a case series comprised of chronic lymphocytic leukaemia and non-Hodgkin lymphoma combined were similar to those for non-Hodgkin lymphoma alone (Delzell et al., 2006). External analyses that compared mortality rates of workers in each cumulative exposure category to the rate in the general population and controlled for age, race and calendar year provided consistent results.

The mortality follow-up of North American synthetic rubber industry workers who had been employed for at least 1 year before 1 January 1992 at any of seven US and one Canadian styrene–butadiene rubber plants was later extended for a further 7 years from 1991 through to the end of 1998 (Sathiakumar et al., 2005) and included 17 924 subjects (15 583 white and 2341 non-white). Work histories provided information on each job held, a description of the work area and job and payroll classification. Analyses of data by work area were restricted to 15 612 subjects who had been employed in styrene–butadiene rubber-related operations in the eight plants and classified into five major work areas: production, maintenance, labour, laboratories and other operations. For six of the eight plants (a total of 14 273 subjects), work histories were sufficiently detailed to permit further specification of three subgroups of work area in production, two in maintenance and two in labour. Vital status was updated through to 1998 and ascertained for 97% of the study group. In total, 570 (3%) subjects were lost to follow-up. A total of 540 586 person–years of observation were accrued. SMRs were calculated using mortality rates of the male population in Ontario (Canada) and in the three US states where the plants were located. The update added 83 401 person–years (18% increase), 1578 deaths (34%), 492 deaths from cancer (44%) and 20 deaths from leukaemia (39%) to those of the previous study. Mortality from all causes was lower than that expected (6237 deaths; SMR, 0.86; 95% CI, 0.84–0.88) as was mortality from all cancers (1608 deaths; SMR, 0.92; 95% CI, 0.88–0.97). Mortality from lymphatic and haematopoietic cancer was slightly elevated (162 deaths; SMR, 1.06; 95% CI, 0.90–1.23); 12 deaths from Hodgkin lymphoma yielded an SMR of 1.11 (95% CI, 0.58–1.95) and 71 deaths from leukaemia represented a modest excess above expectation (SMR, 1.16; 95% CI, 0.91–1.47). Mortality from non-Hodgkin lymphoma was as expected (53 deaths; SMR, 1.00; 95% CI, 0.75–1.30) and that from multiple myeloma was lower than expected (26 deaths; SMR, 0.95; 95% CI, 0.62–1.40). No consistent patterns were observed when mortality was analysed by ever/never hourly paid, by years since hire or by years worked. In the overall study group, the excess mortality from leukaemia was concentrated among men who had been hired in the 1950s (31 deaths; SMR, 1.50; 95% CI, 1.01–2.11). The excess mortality from all leukaemias was highest among hourly paid workers (63 deaths; SMR, 1.23; 95% CI, 0.94–1.57), especially among those who had 20–29 years since hire and ≥ 10 years worked (19 deaths; SMR, 2.58; 95% CI, 1.56–4.03). Analysis by specific histological type was possible for 65 of the 71 leukaemias and for the 1968–98 time period only. For all subjects included in this analysis, the SMR for all forms of leukaemia (65 deaths) was 1.26 (95% CI, 0.97–1.61), that for lymphocytic leukaemia (19 deaths) was 1.28 (95% CI, 0.77–2.00) and that for myelogenous leukaemia (28 deaths) was 1.27 (95% CI, 0.84–1.83). In the SMR analysis restricted to 1968–98, hourly paid workers had an overall SMR for leukaemia (65 deaths) of 1.35 (95% CI, 1.03–1.75), and the excess mortality was particularly marked among those who had 20–29 years since hire and ≥ 10 years worked (SMR, 2.84; 95% CI, 1.68–4.49); a non-significant increase was seen for lymphocytic leukaemia (four deaths; SMR, 2.33; 95% CI, 0.65–5.97), whereas the excess for myelogenous leukaemia was significant (nine deaths; SMR, 3.20; 95% CI, 1.46–6.07) and was stronger for the chronic myelogenous form (six deaths; SMR, 6.55; 95% CI, 2.40–14.26). Analysis by production work areas in the whole cohort, with follow-up between 1944 and 1998, showed an association between mortality from leukaemia and working in the following areas: polymerization (18 deaths; SMR, 2.04; 95% CI, 1.21–3.22), coagulation (10 deaths; SMR, 2.31; 95% CI, 1.11–4.25) and finishing (19 deaths; SMR, 1.56; 95% CI, 0.94–2.44); in maintenance labour (15 deaths; SMR, 2.03; 95% CI, 1.14–3.35); and in laboratories (14 deaths; SMR, 3.26; 95% CI, 1.78–5.46). The excesses of mortality in production were mainly due to chronic lymphatic leukaemia: polymerization (eight deaths; SMR, 4.97; 95% CI, 2.15–9.80), coagulation (five deaths; SMR, 6.07; 95% CI, 1.97–14.17) and finishing (seven deaths; SMR, 3.44; 95% CI, 1.38–7.09), whereas mortality from myelogenous leukaemia was particularly high in maintenance labour (acute, five deaths; SMR, 2.95; 95% CI, 0.96–6.88) and in laboratory workers (total, six deaths; SMR, 3.31; 95% CI, 1.22–7.20; chronic, three deaths; SMR, 5.22; 95% CI, 1.08–15.26).

Cheng et al. (2007) used Cox proportional hazard models on the set of data analysed by Graff et al. (2005) to examine further the exposure–response relations between several butadiene exposure indices and leukaemia (81 decedents), and to assess exposure–response relations between butadiene and all lymphoid neoplasms (120 decedents from lymphatic leukaemia, non-Hodgkin lymphoma, Hodgkin lymphoma and multiple myeloma) and all myeloid neoplasms (56 decedents from myeloid and monocytic leukaemia, myelofibrosis, myelodysplasia, myeloproliferative disorders and polycythemia vera). Cox regression techniques were considered to permit estimation of the exposure–response relations throughout the exposure range, to provide optimal control of confounding by age and to be less affected by intercorrelations among exposure variables. A subset of 488 subjects was excluded because they dropped out from follow-up at ages below that of the youngest leukaemia decedent. Potential confounders for which the analyses controlled included dimethyldithiocarbamate, race, plant, years since hire and year of birth. The butadiene exposure indices used in these analyses were: cumulative exposure in ppm–years, total number of exposures to peaks (> 100 ppm) and average intensities of exposure in parts per million. All three exposure indices were associated positively with the risk for leukaemia. Penalized spline regression indicated that the natural logarithm of the hazard ratio for leukaemia increased in a fairly linear fashion in the exposure range below the 95th percentile for all three indices. Analysis by decile of exposure to butadiene showed an irregular pattern of estimated rate ratios. After controlling for all co-variates, estimated relative rates by decile range of values exhibited a non-monotonic trend that was of borderline statistical significance for ppm–years [p for trend = 0.049], but not for exposure to peaks [p for trend = 0.071] or average intensity of exposure [p for trend = 0.433]. Models that used continuous exposure variables indicated that, for butadiene ppm–years, the exposure–response relation with leukaemia was positive and statistically significant in all but two of the eight models evaluated. After adjustment for all co-variates, the regression coefficient b that estimated the slope of the exposure–response relation was 3.0 × 10⁻⁴ (95% CI, 0.1 × 10⁻⁴−5.8 × 10⁻⁴; p = 0.04) in models that used continuous, untransformed ppm–years, 5.8 × 10⁻⁴ (95% CI, 0.5 × 10⁻⁴−11.1 × 10⁻⁴; p = 0.03) in models that used mean scored deciles and 6.1 × 10⁻⁴ (p = 0.04) in models that used mean scored quintiles. The exposure–response trend for butadiene peaks and leukaemia was positive and statistically significant in all eight models. The regression coefficient b was 5.6 × 10⁻⁵ (95% CI, 0.8 × 10⁻⁵−10.4 × 10⁻⁵; p = 0.02) in models that used continuous, untransformed butadiene peaks and 7.5 × 10⁻⁵ (95% CI, 0.3 × 10⁻⁵−14.7 × 10⁻⁵; p = 0.04) in models that used mean scored deciles of butadiene peaks, after controlling for all covariates. The association of average intensity of exposure to butadiene with leukaemia was statistically significant only in the model that used the square-root transformation of parts per million butadiene. Lagging of exposure had a small impact on the value of the coefficients for the three exposure variables. Lymphoid neoplasms were associated with the ppm–years exposure index and myeloid neoplasms with exposure to peaks. However, neither of the relations was statistically significant after control for covariates. Cumulative exposure to dimethyldithiocarbamate, when treated as a continuous variable, was not associated with leukaemia in any of the models. The risk estimates for each quartile of cumulative exposure to dimethyldithiocarbamate were significantly increased, even after adjustment for cumulative exposure to butadiene, without, however, a monotonic trend. Relative rate estimates were, respectively, 2.4 (95% CI, 1.2–5.0), 2.9 (95% CI, 1.4–5.8), 4.5 (95% CI, 2.2–8.9) and 2.1 (95% CI, 1.0–4.4) [p for trend = 0.458].

[The Working Group noted that, given the strong correlation between original and revised estimates (Spearman's r = 0.9) reported, the validation of the exposure estimates in one plant during 1977–91 (Sathiakumar & Delzell, 2007) showed that estimates from the most recent measurements were in general very close to measured exposures, especially for the styrene–butadiene rubber-related job titles. The ranking of jobs had hardly changed from the first estimates (Macaluso et al., 1996) to the adjusted estimates (Macaluso et al., 2004). However, it is uncertain to what extent this validation can be extrapolated to the other plants.]

2.2.3. Other rubber production

Bond et al. (1992) reported a mortality study of 2904 male workers engaged in the development and manufacture of styrene-based products, including styrene–butadiene latex, who were potentially exposed to styrene and related chemicals [including butadiene] for at least 1 year between 1937 and 1971. The number of person–years of follow-up during 1970–86 for workers in this production was 11 754. In comparison with USA mortality rates, the SMR for all causes of death among styrene–butadiene latex workers was 0.9, based on 82 deaths. A total of 13 cancers were observed (22.0 expected; SMR, 0.6), and no site had an SMR that exceeded unity. One death from lymphatic and haematopoietic cancers occurred, which was due to leukaemia (ICD-8 204–207) versus 0.9 expected. [The Working Group noted the limited information relating to exposure to butadiene.]

3.1.1. Mouse

Groups of 50 male and 50 female B6C3F₁ mice, 8–9 weeks of age, were exposed by whole-body inhalation to 625 or 1250 ppm [1380 or 2760 mg/m³] butadiene (minimum purity, > 98.9%) for 6 h per day on 5 days per week for 60 weeks (males) or 61 weeks (females). Equal numbers of animals were sham-exposed and served as controls. The study was terminated after 61 weeks because of a high incidence of lethal neoplasms in the exposed animals. Deaths were mainly due to malignant lymphomas. The numbers of survivors at 61 weeks were: males — 49/50 control, 11/50 low-dose and 7/50 high-dose; females — 46/50 control, 15/50 low-dose and 30/50 high-dose. As shown in Table 19, butadiene induced haemangiosarcomas that originated in the heart and metastasized to various organs. The incidence of haemangiosarcomas of the heart in historical controls was extremely low (1/2372 males, 1/2443 females). Other types of neoplasm for which the incidence was significantly increased (Fisher's exact test) in animals of each sex were malignant lymphomas, alveolar/bronchiolar adenomas or carcinomas of the lung and papillomas or carcinomas of the forestomach. Tumours that occurred with significantly increased incidence in females only included hepatocellular adenoma or carcinoma of the liver: 0/50 control, 2/47 (p = 0.232) low-dose and 5/49 (p = 0.027) high-dose; acinar-cell carcinoma of the mammary gland: 0/50 control, 2/49 low-dose and 6/49 (p = 0.012) high-dose; and granulosa-cell tumours of the ovary: 0/49 control, 6/45 (p = 0.01) low-dose and 12/48 (p < 0.001) high-dose. Gliomas were observed in the brain of one male mouse exposed to 1250 ppm and in two male mice exposed to 625 ppm butadiene (National Toxicology Program, 1984; Huff et al., 1985).

Table 19

Incidence of tumours in B6C3F₁ mice exposed to butadiene by inhalation at 625 and 1250 ppm for 61 weeks.

Because of the reduced survival of mice in the initial study, further studies were conducted at lower exposure concentrations. In one study, groups of 70–90 male and 70–90 female B6C3F₁ mice, 6.5 weeks of age, were exposed by whole-body inhalation to 0, 6.25, 20, 62.5, 200 or 625 ppm [0, 14, 44, 138, 440 or 1380 mg/m³] butadiene (purity, > 99%) for 6 h per day on 5 days per week for up to 2 years. Up to 10 animals per group were killed and evaluated after 40 and 65 weeks of exposure. Survival was significantly reduced (p < 0.05) in all groups of mice exposed to 20 ppm or higher; terminal survivors were: males — 35/70 control, 39/70 at 6.25 ppm, 24/70 at 20 ppm, 22/70 at 62.5 ppm, 3/70 at 200 ppm and 0/90 at 625 ppm; females — 37/70 control, 33/70 at 6.25 ppm, 24/70 at 20 ppm, 11/70 at 62.5 ppm, 0/70 at 200 ppm and 0/90 at 625 ppm. As shown in Table 20, exposure to butadiene produced increases in both sexes in the incidence of lymphomas, heart haemangiosarcomas, lung alveolar/bronchiolar adenomas and carcinomas, forestomach papillomas and carcinomas, Harderian gland adenomas and adenocarcinomas and hepatocellular adenomas and carcinomas. The incidence of mammary gland adenocarcinomas and benign and malignant ovarian granulosa-cell tumours was increased in females. Lymphocytic lymphomas were seen as early as after 23 weeks of exposure and were the principal cause of death in male and female mice exposed to 625 ppm butadiene (Melnick et al., 1990; National Toxicology Program, 1993).

Table 20

Survival and incidence of tumours in mice exposed to butadiene by inhalation for up to 2 years.

In the same study, a stop-exposure experiment was conducted in which groups of 50 male B6C3F₁ mice, 6.5 weeks of age, were exposed to butadiene (purity, > 99%) by whole-body inhalation for 6 h per day on 5 days per week at concentrations of 200 ppm [440 mg/m³] for 40 weeks, 312 ppm [690 mg/m³] for 52 weeks, 625 ppm [1380 mg/m³] for 13 weeks or 625 ppm [1380 mg/m³] for 26 weeks. The multiple of the exposure concentration and duration of exposure (ppm–weeks) was approximately 8000 ppm–weeks or 16 000 ppm–weeks for the exposure groups. After the exposures were terminated, the animals were placed in control chambers for up to 104 weeks after the beginning of treatment. A group of 70 males served as chamber controls (0 ppm). [The Working Group noted that this was the same control group as that used in the experiment described in the above paragraph.] Survival was reduced (p < 0.05) in all exposed groups; the numbers of survivors at the end of the study were 35 controls, nine exposed to 200 ppm, one exposed to 312 ppm, five exposed to 625 ppm for 13 weeks and none exposed to 625 ppm for 26 weeks. As shown in Table 21, exposure to butadiene produced increases in the incidence of lymphomas, heart haemangiosarcomas, lung alveolar/bronchiolar adenomas and carcinomas, forestomach papillomas and carcinomas, Harderian gland adenomas and adenocarcinomas, preputial gland carcinomas and kidney tubular adenomas. This exposure protocol revealed additional tumour sites in males (preputial gland and renal cortex). The incidence of lymphocytic lymphoma was greater after exposure to higher concentrations of butadiene for a short time than after exposure to lower concentrations for an extended period (Melnick et al., 1990; National Toxicology Program, 1993). Brain neoplasms including two neuroblastomas and three gliomas were observed in mice exposed to 625 ppm for 13 or 26 weeks in the stop-exposure study (National Toxicology Program, 1993). Brain neoplasms are rare and had never been seen in historical National Toxicology Program controls at the time of the study. [The Working Group noted that brain tumours are rare in mice and were observed, although at a low incidence, in both National Toxicology Program bioassays (National Toxicology Program, 1984, 1993).]

Table 21

Survival and incidence of tumours in male mice exposed to butadiene in stop-exposure studies.

Follow-up studies were completed to test the hypothesis that the high incidence of lymphocytic lymphomas in mice exposed to concentrations of 200 ppm butadiene or higher was at least partially dependent on the activation of an endogenous retrovirus in the strain of mouse used in the study. Groups of 60 male B6C3F₁ and 60 male NIH Swiss mice, 4–6 weeks of age, were exposed to 0 or 1250 ppm [2760 mg/m³] butadiene (> 99.5% pure) by whole-body inhalation for 6 h per day on 5 days per week for 52 weeks. An additional group of 50 male B6C3F₁ mice was exposed similarly to butadiene for 12 weeks and held until termination of the experiment at 52 weeks. The incidence of thymic lymphomas was 1/60 control, 10/48 exposed for 12 weeks and 34/60 exposed for 52 weeks in B6C3F₁ mice and 8/57 in NIH Swiss mice exposed for 52 weeks. Haemangiosarcomas of the heart were observed in 5/60 B6C3F₁ mice and 1/57 NIH Swiss mice exposed for 52 weeks. Exposure of B6C3F₁ mice to 1250 ppm of butadiene for 6 h per day on 5 days per week for 3–21 weeks greatly increased the quantity of ecotropic retrovirus recovered from bone marrow, thymus and spleen. This was not the case for NIH Swiss mice in which proviral ecotropic sequences are truncated and the virus is not expressed. The authors suggested that the lack of retroviruses provides resistance to the induction of the lymphomas by butadiene (Irons et al., 1987a, 1989; Irons, 1990). [The Working Group noted that Swiss mice are less sensitive to the induction of haemangiosarcomas of the heart and lymphomas and that genetic factors may contribute to the development of these tumours.]

One study addressed the question of whether a single, high level of exposure to butadiene was sufficient to induce neoplasia. Groups of 60 male and 60 female B6C3F₁ mice, 8–10 weeks of age, were exposed by whole-body inhalation for a single 2-h period to 0, 1000, 5000 or 10 000 ppm [0, 2200, 11 000 or 22 000 mg/m³] butadiene [purity unspecified]. The mice were then held for 2 years, at which time all survivors were killed and tissues and organs were examined histopathologically. Survival, weight gains and tumour incidence in exposed mice were not affected by exposure to butadiene (survival: males — 28/60 control, 34/60 low-dose, 44/60 mid-dose and 34/60 high-dose; females — 45/60, 36/60, 38/60 and 48/60, respectively) (Bucher et al., 1993).

3.1.2. Rat

Groups of 100 male and 100 female Sprague-Dawley rats, 4–5 weeks of age, were exposed by whole-body inhalation to 0, 1000 or 8000 ppm [0, 2200 or 17 600 mg/m³] butadiene (minimal purity, 99.2%) for 6 h per day on 5 days per week for 111 weeks (males) or 105 weeks (females). Survival was reduced in low- (p < 0.05) and high-dose (p < 0.01) females and in high-dose males (p < 0.05); the numbers of survivors were: males — 45 control, 50 low-dose and 32 high-dose; females — 46 control, 32 low-dose and 24 high-dose. Tumours that occurred at a significantly increased incidence in males were pancreatic exocrine adenomas (control, 3/100; low-dose, 1/100; and high-dose, 10/100, p ≤ 0.001) and interstitial-cell tumours of the testis (control, 0/100; low-dose, 3/100; and high-dose, 8/100; p for trend ≤ 0.001). Those that occurred at a significantly increased incidence in females were follicular-cell adenomas of the thyroid gland (control, 0/100; low-dose, 2/100; and high-dose, 10/100; p for trend ≤ 0.01), sarcomas of the uterus (control, 1/100; low-dose, 4/100; and high-dose, 5/100; p for trend ≤ 0.05), carcinomas of the Zymbal gland (control, 0/100; low-dose, 0/100; and high-dose, 4/100; p for trend ≤ 0.01) and benign and malignant mammary tumours (control, 50/100; low-dose, 79/100; and high-dose, 81/100; p for trend ≤ 0.001). Mammary adenocarcinomas were found in 18/100 control, 15/100 low-dose and 26/100 high-dose female rats (Owen et al., 1987; Owen & Glaister, 1990).

3.2.1. 1,2-Epoxy-3-butene (epoxybutene)

3.2.2. 1,2:3,4-Diepoxybutane (diepoxybutane)

4.1.1. Humans

4.1.2. Experimental systems

At the beginning of each of the following sections, brief highlights on the disposition of butadiene that were summarized previously (IARC, 1999) are presented. The reader should refer to that monograph for citations and more complete descriptions of the relevant studies. These highlights are followed by information that has been published since that time. For this and other sections, diepoxybutane refers to the (±) racemic mixture of diastereomers, unless otherwise specified.

(b) Metabolites in vitro (see Figure 1)

Butadiene is oxidized initially to epoxybutene by nicotinamide adenine dinucleotide phosphate (NADPH)-dependent metabolism that primarily involves CYP2E1 and CYP2A6. This metabolic pathway has been measured in microsomal fractions obtained from the livers and lungs of mice, rats and humans and from the kidney and testis of rats and mice (IARC, 1999). At low concentrations of butadiene, metabolism via CYP2E1 predominates. Liver microsomes from mice, rats and humans can also oxidize epoxybutene to diepoxybutane. Kinetic parameters (maximum velocity [V_max] and Michaelis-Menten constant [K_m]) indicated faster rates of oxidative metabolism of butadiene and epoxybutene in microsomes from mice than in those from rats, while human liver microsomes showed a wide range of activites (Csanady et al., 1992; Duescher & Elfarra, 1994), but only a small number of human liver and lung samples were analysed. Mouse, rat and human microsomes can convert epoxybutene to both (±)-diepoxybutane and meso-diepoxybutane. The latter isomer is preferentially hydrolysed in rat and human liver microsomes (Krause & Elfarra, 1997). Epoxybutene and diepoxybutane are eliminated by microsomal EH (mEH) and by cytosolic GST activities. In-vitro kinetics of EH- and GST-catalysed metabolism of epoxybutene and diepoxybutane in microsomal and cytosolic fractions from livers and lungs of mice, rats and humans have also been reported (IARC, 1999). GST activities were highest in mouse and lowest in human cytosol; EH activities were highest in human and lowest in mouse microsomes. Bone-marrow cells of mice and humans can also oxidize butadiene to epoxybutene by a myeloperoxidase-catalysed reaction. A third epoxide intermediate of butadiene metabolism, epoxybutanediol, can be formed by the oxidation of butenediol (the product of epoxybutene hydrolysis) or by partial hydrolysis of diepoxybutane.

Hepatocytes isolated from male B6C3F₁ mice and Sprague-Dawley rats were incubated with epoxybutene and evaluated for the formation of diepoxybutane, butenediol and GSH conjugates at 5-min intervals during incubations of up to 45 min (Kemper et al., 2001). The results are shown in Table 22. Cumulative levels of combined racemic and meso-diepoxybutane (oxidation of epoxybutene) were higher in mouse than in rat hepatocyte cultures, whereas the cumulative levels of butenediol (hydrolysis of epoxybutene) were higher in the rat hepatocyte cultures. GSH conjugates were present at similar levels in both rat and mouse hepatocyte cultures.

Table 22

Area under the concentration versus time curve (nmol·min/10⁶ cells) for epoxybutene-derived metabolites in isolated mouse and rat hepatocytes during 45-min incubations with epoxybutene.

Hepatic microsomes pooled from cynomolgus monkeys or humans exposed to butadiene (monkey microsomes: 46 ppm, 290 ppm or 28 000 ppm [102, 641 or 61 880 mg/m³]; human microsomes: 45 ppm, 450 ppm or 36 000 ppm [99, 995 or 79 560 mg/m³]) produced epoxybutene at similar rates (Dahl & Henderson, 2000). Oxidation of epoxybutene to diepoxybutane, which is competitively inhibited by butadiene, was slightly faster in monkey than in human microsomes.

(c) Metabolites in vivo (see Figure 1)

The identification of epoxybutene in exhaled breath, blood and multiple organs of rats or mice exposed to butadiene indicates the systemic availability of this metabolic intermediate. At equivalent exposure concentrations of butadiene, tissue levels of epoxybutene and diepoxybutane were higher in mice than in rats (IARC, 1999).

¹⁴C-Labelled epoxybutene was administered as a single intraperitoneal injection to male Sprague-Dawley rats and B6C3F₁ mice at doses of 1, 5, 20 and 50 mg/kg bw; urine and faeces were collected up to 48 h after treatment (Richardson et al., 1998). Approximately 50% of the administered radioactivity was excreted in the urine and 2–5% in the faeces. Urinary metabolites identified in this study are listed in Table 23. Because the percentage of urinary radioactivity recovered for each metabolite did not differ in relation to the administered dose, mean values across the dose groups are shown. Both species preferentially metabolized epoxybutene by direct conjugation with GSH. In rats, 48–64% of the urinary radioactivity was derived from the direct reaction of epoxybutene with GSH and 14–25% was derived from hydrolysis of epoxybutene. In mice, 60–72% of the urinary radioactivity was derived from the direct reaction of epoxybutene with GSH and 6–10% was derived from hydrolysis of epoxybutene. No metabolites of diepoxybutane were detected in the urine of either species.

Table 23

Urinary metabolites in rats and mice given a single intraperitoneal injection of 1, 5, 20 or 50 mg/kg bw [4-¹⁴C]epoxybutene.

Male Sprague-Dawley rats and B6C3F₁ mice were exposed by nose-only inhalation to 200 ppm [442 mg/m³] [2,3-¹⁴C]butadiene for 6 h; radioactivity in urine, faeces, exhaled volatiles and [¹⁴C]carbon dioxide was measured during and up to 42 h after exposure (Richardson et al., 1999). Total uptake of butadiene was 0.19 mmol/kg bw (10.3 mg/kg) in rats and 0.38 mmol/kg bw (20.5 mg/kg) in mice. In rats, 40% of the recovered radioactivity was exhaled as [¹⁴C]carbon dioxide, 42% was excreted in urine, 8% was excreted in faeces and 9% remained in the carcass. In mice, 6% of the recovered radioactivity was exhaled as [¹⁴C]carbon dioxide, 71% was excreted in urine, 11% was excreted in faeces and 13% remained in the carcass. Because the position of the radiolabel in this study was on carbons 2 and 3, the formation of [¹⁴C]carbon dioxide is a result of the loss of two carbon atoms from butadiene. Urinary metabolites identified in this study are listed in Table 24. Metabolites arising from the direct reaction of epoxybutene with GSH accounted for 8% of the metabolized dose in rats and 16% of the metabolized dose in mice. In contrast to the study of epoxybutene metabolism (Richardson et al., 1998), the study on butadiene metabolism identified trihydroxybutyl mercapturic acids in the urine of exposed rats and mice.

Table 24

Urinary metabolites in rats and mice exposed to 200 ppm [¹⁴C]butadiene for 6 h.

Urinary metabolites were also identified in male Sprague-Dawley rats and B6C3F₁ mice exposed by nose-only inhalation to 1, 5 or 20 ppm [2.21, 11 or 44 mg/m³] [2,3-¹⁴C]butadiene for 6 h (Booth et al., 2004a). This study did not include measurements of radioactivity in faeces, exhaled volatiles, carbon dioxide or the carcass. Urine was collected over an 18-h period after exposure but not during exposure. The profiles of urinary metabolites were somewhat similar to those reported by Richardson et al. (1999).

Four mercapturic acids derived from HMVK were identified in the urine from male B6C3F₁ mice and Sprague-Dawley rats given a single intraperitoneal injection of 25, 62.5, 125 or 200 mg/kg bw butanediol (Sprague & Elfarra, 2004). The metabolites were N-acetyl-l-cystein-S-yl conjugates of 1,2-dihydroxybutane, 1-hydroxy-2-butanone, propanol and propionic acid. HMVK may be formed by ADH- or CYP-mediated oxidation of butenediol. These four HMVK-derived metabolites accounted for 7 and 11% of the total radioactivity recovered from rats and mice exposed to butadiene, respectively (Richardson et al., 1999).

Trace amounts of 1-hydroxy-2-butanone, crotonic acid, propionic acid and 2-ketobutyric acid were detected in the urine of male B6C3F₁ mice and Sprague-Dawley rats administered butanediol by intraperitoneal injection (mice: 125 or 250 mg/kg bw; rats: 62.5, 125, 200 or 250 mg/kg bw) (Sprague & Elfarra, 2003). The combined concentration of these metabolites was less than 1% of the administered dose of butenediol; these low levels may be due to further metabolism of the carboxylic acids prior to their excretion in the urine. The detection of these metabolites suggests the possibility that toxic intermediary metabolites, such as crotonaldehyde, may be formed from butenediol.

(d) Haemoglobin adducts

Epoxybutene can form two diastomeric pairs of adducts at the N-terminal valine of haemoglobin termed MHbVal: N-(2-hydroxy-3-buten-1-yl)valine and N-(1-hydroxy-3-buten-2-yl)valine (IARC, 1999). Levels of these adducts formed in rats and mice are dependent on exposure concentrations of butadiene, duration of exposure and the turnover rates of red blood cells.

Much work has been carried out on the use of haemoglobin adducts as biomarkers of internal levels of the epoxide intermediates of butadiene metabolism. Epoxybutene, diepoxybutane and epoxybutanediol can react with the N-terminal valine to form MHbVal, PyrVal and THbVal adducts, respectively (Boysen et al., 2007). In addition, THbVal may also be formed by the reaction of haemoglobin with diepoxybutane and subsequent hydrolysis of the remaining epoxide group.

THbVal adduct levels in haemoglobin isolated from erythrocytes of male Sprague-Dawley rats and B6C3F₁ mice exposed by nose-only inhalation to 1, 5 or 20 ppm [2.2, 11 or 44 mg/m³] [2,3-¹⁴C]butadiene for 6 h per day for 5 days were 80, 179 and 512 pmol/g haemoglobin for rats, respectively, and 143, 351 and 1100 pmol/g haemoglobin for mice, respectively (Booth et al., 2004a). However, the level of radioactivity was approximately 1.3-fold higher in rat than in mouse haemoglobin. The measured levels of THbVal adduct represented only about 1–2% of the total radioactive adducts in the haemoglobin, which indicates that multiple alkylation sites within haemoglobin may be modified by butadiene metabolites. After incubation of epoxybutane with erythrocytes, Moll et al. (2000) detected 10 epoxybutane adducts on each of the α- and β-globin chains.

A specific ring-closed adduct, the pyrolidine adduct PyrVal, can be formed by the reaction of haemoglobin with diepoxybutane (Fred et al., 2004). This adduct has been measured after in-vitro reactions of rat and mouse erythrocyte haemolysates with diepoxybutane and after intraperitoneal injections of diepoxybutane into Fischer 344 rats and C57/BL mice. The levels of PyrVal adduct per administered dose were similar in rat and mouse haemoglobin after in-vitro or in-vivo exposure to diepoxybutane. In addition, THbVal adducts were measured in rats and mice given intraperitoneal injections of diepoxybutane or epoxybutanediol. Levels of this adduct were similar in mice and rats dosed with epoxybutanediol; however, after treatment with diepoxybutane, levels of THbVal adducts were three to four times higher in rats.

Several haemoglobin adducts were measured in rats and mice after inhalation exposure to butadiene (Boysen et al., 2004, 2007). Female B6C3F₁ mice were exposed to 3, 62.5 or 1250 ppm [6.6, 138 or 3453 mg/m³] butadiene for 6 h per day for 2 weeks, while female Fischer 344 rats were exposed to 3 or 62.5 ppm [6.6 or 138 mg/m³] for 10 days. At equivalent butadiene exposures, the levels of MHbVal and PyrVal were higher in mice than in rats, while levels of THbVal were similar in these species (Table 25). The formation of each of these adducts in mice and rats was more efficient (pmol adduct/g haemoglobin per ppm butadiene) at 3 ppm butadiene than at higher exposure concentrations. The major adduct formed in mice and rats is THbVal. In rats, formation of THbVal reaches a plateau at exposures to butadiene above 62.5 ppm (Boysen et al., 2007).

Table 25

N-Terminal haemoglobin adducts (pmol/g haemoglobin and adducts/ppm butadiene) in mice and rats exposed to butadiene (BD) for 10 days.

The levels of THbVal adducts in female B6C3F₁ mice exposed by inhalation to 6, 18 or 36 ppm butanediol for 6 h per day on 5 days per week for 4 weeks were approximately twofold higher than those in female Fischer 344 rats exposed to the same concentrations of this metabolite. The similarity in the shape of the dose–response curves for the formation of these adducts and the induction of hypoxanthine–guanine phosphoribosyl transferase gene (Hprt) mutant frequency in splenic T cells from mice and rats exposed to butenediol suggests that epoxybutanediol (the product of butenediol epoxidation) may play a significant role in the mutagenicity of butadiene (Powley et al., 2005).

4.2.1. Humans

The HPRT variant frequency in lymphocytes as well as the concentration of the urinary metabolite of butadiene, DHBMA, was examined in 49 workers in a styrene–butadiene rubber production facility in Texas, USA (Ammenheuser et al., 2001). The study included 24 subjects who had high exposure and 25 who had low exposure for HPRT variant frequency analysis, 22 subjects who had high exposure and 24 who had low exposure for dosimeter measurements and 24 who had high exposure and 23 who had low exposure for urine analyses. The mean value of exposure to butadiene was 1.48 ± 0.37 ppm [3.27 ± 0.82 mg/m³] (mean ± standard error [SE]) for the high-exposure group and 0.15 ± 0.02 ppm [0.33 ± 0.04 mg/m³] for the low-exposure group. The frequency of HPRT variants was 6.66 ± 1.4 × 10⁻⁶ (mean ± SE) for the high-exposure group and 2.10 ± 0.2 × 10⁻⁶ for the low-exposure group (p < 0.0002). When smoking status was considered, the frequency of HPRT variants was 6.8 ± 1.2 × 10⁻⁶ for 19 nonsmokers in the high-exposure group and 1.8 ± 0.2 × 10⁻⁶ for 20 nonsmokers in the low-exposure group (p < 0.0005). The concentration of DHBMA was 2046 ± 348 ng/mg creatinine in the urine of the high-exposure group compared with 585 ± 98 ng/mg creatinine in the urine of the low-exposure group (p < 0.0005).

Another study was conducted in the same styrene–butadiene facility in 1998 (Ward et al., 2001). The frequency of HPRT variants in lymphocytes as well as the concentration of the urinary metabolite of butadiene, DHBMA, were examined in 37 workers (22 who had high exposure and 15 who had low exposure). The mean value of exposure to butadiene was 1.71 ± 0.54 ppm [3.78 ± 1.2 mg/m³] (mean ± SE) for the high-exposure group and 0.07 ± 0.03 ppm [0.15 ± 0.07 mg/m³] for the low-exposure group. The frequency of HPRT variants was 10.67 ± 1.51 × 10⁻⁶ (mean ± SE) for the high-exposure group and 3.54 ± 0.61 × 10⁻⁶ for the low-exposure group (p < 0.001). The concentration of the metabolite was 378 ± 53 ng/mg creatinine in the urine of the high-exposure group compared with 271 ± 50 ng/mg creatinine in the urine of the low-exposure group. Unlike the previous study, this difference was not significant (p > 0.05).

A subset of the workers from the Texas facility was analysed for their frequency of HPRT mutants (Ma et al, 2000). HPRT mutants were analysed by a multiplex polymerase chain reaction (PCR) and the frequency of large deletions in butadiene-exposed workers (17.5%; 25/143) was significantly higher than that in control subjects (9.7%; 21/217; p < 0.05). This increase in large deletions was due primarily to an increase in multiple exon deletions (p < 0.05). When the HPRT mutants were analysed for cDNA sequence mutations, the majority of the mutations observed in both exposure groups were single-base substitutions. However, the overall distribution of the types of mutation was significantly different between the two groups (p < 0.05). A non-significant increase in mutations at AT sites was observed in butadiene-exposed workers (46%) compared with the control group (39%), and the proportion of A:T→T:A transversions was also increased in the butadiene-exposed group (16%) compared with the control group (8%; p = 0.25). Three new mutable sites were identified at positions 116, 370 and 410. The frequency of −1 frame shift mutations was significantly higher (11%; p < 0.05) in exposed workers than in the controls (2%). Polymorphisms in the mEH gene may play a significant role in the sensitivity of humans to the genotoxic effects of butadiene. In workers exposed to > 150 ppb [331.5 µg/m³] butadiene, individuals who had at least one polymorphic mEH His allele had a significant (p < 0.001) threefold increase in the frequency of HPRT variant (mutant) lymphocytes compared with individuals who had the Tyr/Tyr genotype (Abdel-Rahman et al., 2001).

Blood samples from 19 exposed (butadiene monomer production unit) and 19 control (heat production unit) workers at a petrochemical company in the Czech Republic were analysed for chromosomal aberrations and sister chromatid exchange (Šrám et al., 1998). The median exposure concentration of butadiene was 0.53 mg/m³ [0.24 ppm] for the exposed group and 0.013 mg/m³ [0.006 ppm] for the control group. A significant increase in the percentage of aberrant cells was observed in the exposed group (3.11 ± 1.33 versus 2.03 ± 1.01%; p < 0.01). When smoking status was considered in the exposed group, no difference in the percentage of aberrant cells was observed between smokers (3.11 ± 1.50) and nonsmokers (3.10 ± 1.24), although nonsmokers were exposed to nearly three times more butadiene (1.73 mg/m³) than smokers (0.53 mg/m³). There was also a significant increase in the frequencies of sister chromatid exchange per cell in the exposed compared with the control groups (6.96 ± 1.51 and 4.87 ± 1.11 [mean ± SD], respectively; p < 0.001).

A comparison of conventional cytogenetic analyses and fluorescence in-situ hybridization (FISH) was conducted among 82 workers in a chemical plant in the Czech Republic (Šrám et al., 2004). Twenty-three subjects worked in monomer production and were exposed to 0.642 mg/m³ [0.3 ppm] butadiene, 34 worked in polymer production (1.794 mg/m³ [0.812 ppm] butadiene) and 25 matched controls (0.023 mg/m³ [0.01 ppm] butadiene) worked in administration. Using both methods, no significant differences in chromosomal aberration frequency were detected between any of the groups. When subjects with suspected clonality were excluded, re-analysis of the data by FISH found a significant difference between the polymer production workers (2.73 ± 1.51 genomic frequencies of stable chromosomal exchanges) and the monomer production and control groups (1.72 ± 1.14 and 2.06 ± 1.31 genomic frequencies of stable chromosomal exchanges, respectively).

A study that used several different biomarkers to determine the effects of exposure to butadiene on workers was undertaken in the Czech Republic (Albertini et al., 2001). Eighty-three workers from the same industrial site were divided into three groups: controls (25 from administration), and monomer (24) and polymer (34) workers. Polymer workers typically had levels of exposure (1.76 ± 4.69 mg/m³ [0.8 ppm]) that were significantly higher than those of the monomer workers (0.64 ± 2.06 mg/m³ [0.29 ppm]) and controls (0.3 ± 0.03 mg/m³ [0.14 ppm]). Analysis by autoradiography of HPRT mutations in 49 workers showed a significant difference between the control (10.75 ± 6.11 × 10⁻⁶) and the monomer (5.73 ± 4.72 × 10⁻⁶) and the polymer workers (6.48 ± 4.77 × 10⁻⁶). This trend contradicts that expected. Analysis by a cloning assay of HPRT mutations in 75 workers showed no significant differences between the groups (controls, 13.00 ± 8.1 × 10⁻⁶; monomer workers, 10.69 ± 5.4 × 10⁻⁶; polymer workers, 18.83 ± 17.41 × 10⁻⁶). Assays for sister chromatid exchange were completed for 73 study subjects. No significant differences were observed between the groups. Chromosomal aberrations were analysed by traditional methods in 82 workers. The mean percentages of cells with aberrations were 1.56, 1.52 and 1.54% for the control, monomer and polymer groups, respectively. The mean number of chromosomal breaks per cell was also similar between the groups. Chromosomal changes analysed by FISH also showed no significant differences. [The Working Group noted the very high levels of butadiene-derived metabolites in the urine of controls.]

A Health Effects Institute Report summarized the study of workers in the Czech Republic (Albertini et al., 2003a). The conclusions on the genotoxicity studies were that ‘none of these measures showed positive responses at exposure levels encountered in this study’. HPRT mutations analysed by cloning or autoradiographic techniques showed no effect of butadiene, and mutation spectra were not significantly different between exposed and unexposed workers. HPRT mutation frequencies were also unrelated to the metabolic genotypes examined. When smoking status was considered, there was no significant effect of smoking and no significant exposure group-by-smoking interaction on HPRT mutation frequencies. Chromosomal aberrations analysed by traditional or FISH methods and sister chromatid exchange were also unaffected by exposure to butadiene. The results of the study by Šrám et al. (1998) were in conflict with these conclusions and showed increased chromosomal aberrations.

Further molecular epidemiological analysis of butadiene-exposed workers in the Czech Republic was conducted to determine whether any gender differences in the response to butadiene existed (Albertini et al., 2007). The average level of exposure to butadiene (mg/m³) was 0.397 ± 0.502 (mean ± SD) [0.18 ppm] for the 23 women and 0.808 ± 1.646 [0.37 ppm] for the 30 men. Thus, exposed male workers had a significantly higher level of exposure than female workers in this study. It should be noted that exposed male workers also had higher levels of exposure to styrene than female workers and that exposed female workers had higher levels of exposure to toluene and benzene than the exposed male workers. Although urinary concentrations of the mercapturic acid metabolites of butadiene, DHBMA and MHBMA, were higher in butadiene-exposed women than in female controls, the differences were not significant. The levels of both DHBMA and MHBMA (µg/L) were significantly increased in male butadiene-exposed subjects (854.1 ± 567.0 and 47.9 ± 44.3, respectively) compared with male controls (512.8 ± 272.1 and 14.9 ± 10.3, respectively; p < 0.05). Thus, significantly higher concentrations of metabolites occurred in men than in women in both the control and exposed groups. HPRT mutations measured by the T-cell assay did not differ significantly between exposed and control groups of either sex. No significant associations between exposure to butadiene and sister chromatid exchange or chromosomal aberrations were detected in either sex. Effects of genotype were also examined. In this study, GSTT1-null workers showed a significantly slower rise in the rate of MHBMA excretion (p < 0.05). In addition, individuals with the EH genotype that specifies low activity showed a significantly higher rise in urinary MHBMA/(DHBMA + MHBMA) ratios with increasing exposure to butadiene.

Zhang et al. (2004) measured chromosomal changes in the peripheral blood lymphocytes of 39 butadiene polymer production workers and 38 unexposed controls in Yanshan, China. The median exposure level for the butadiene-exposed workers as a 6-h TWA was 2 ppm [4.42 mg/m³] whereas the control group had a median level of 0 ppm. Tobacco use was controlled for by including a similar percentage of smokers with similar pack–years of smoking in both the control and exposed groups. No significant numerical or structural chromosomal changes were detected using FISH with probes for chromosomes 1, 7, 8 or 12. GSTT1 and GSTM1 genotypes had no significant effect on the frequency of hyperdiploidy of the above chromosomes or on the frequency of structural changes of chromosomes 8 and 12 in the butadiene-exposed group. The EH EPHX1 Y113H polymorphism had no effect on chromosomal damage or HPRT mutant frequency in either the exposed or control workers. However, workers with the histidine arginine HR or RR allele of the EPHX1 H139R polymorphism had increased levels of hyperdiploidy of chromosomes 1, 7 and 8. Overall, predicted EPHX1 activity did not influence genetic damage at low occupational exposures to butadiene. Further analysis at this plant of 41 workers exposed to the same median levels of butadiene (2 ppm) and 38 controls was undertaken (Hayes et al., 2001). No differences were observed in the Glycophorin A assay between the exposed workers and the controls. The mutation frequency of HPRT measured with the T-cell cloning assay was also not significantly different and no significant increase was detected in sister chromatid exchange frequency.

Another study at a tyre plant in the Slovak Republic examined 110 workers, who were exposed to several xenobiotics of which butadiene was the most prominent, for markers of genotoxicity in relation to several polymorphisms (Vodicka et al., 2004). The workers were divided into three groups: high exposure (butadiene concentration, 2.6 ± 0.2 mg/m³ [1.18 ppm]), low exposure (butadiene concentration, 2.3 ± 2.2 mg/m³ [1.04 ppm]) and no exposure (trace amounts). The frequencies of total chromosomal aberrations were significantly lower (1.3 ± 1.3; p < 0.01) in the low-exposure group compared with the high- and no-exposure groups (2.2 ± 1.4 and 2.3 ± 1.1, respectively). No significant differences were observed in DNA single-strand breaks or in single-strand break endo III-sensitive site frequencies. A non-significant twofold higher rate of DNA repair was found in the high-exposure group (0.6 ± 0.6 single-strand breaks/10⁹ daltons) compared with the low- and no-exposure groups (0.3 ± 0.3 and 0.3 ± 0.4 single-strand breaks/10⁹ daltons, respectively). A weak but non-significant association was found between CYP2E1 expression in peripheral blood lymphocytes and the frequency of chromosomal aberrations (r = 0.298; p = 0.097). In all individuals, assessment of several genetic polymorphisms suggested that individuals who had low EPHX1-activity genotypes had the highest level of chromosomal aberrations. Significantly lower frequencies of chromosomal aberration (p = 0.024) were detected in individuals who had the variant CC genotype associated with the XPD exon 23.

A study of 27 healthy male Caucasian workers exposed to butadiene and 26 matched controls from an Italian petrochemical plant analysed genotoxic effects (Lovreglio et al., 2006). The mean exposure to butadiene was 6.4 ± 14.0 µg/m³ [0.003 ppm] for the butadiene-exposed workers which was significantly different from that of the controls (0.8 ± 1.1 µg/m³ [0.0004 ppm]; p < 0.001). No significant differences were observed in sister chromatid exchange, the percentage of cells with a high frequency of sister chromatid exchange, chromosomal aberrations or proliferation index in the peripheral blood lymphocytes of these two groups. When subjects were classified according to smoking status, a significant increase was observed in the mean frequency of sister chromatid exchange in smokers (6.6 ± 1.2) compared with nonsmokers (5.5 ± 0.8; p = 0.001). Exposure to butadiene was also higher in smokers than in nonsmokers but this was not statistically significant (p = 0.3).

4.2.2. Experimental systems

(a) Butadiene (see Table 26 for details and references)

Table 26

Genetic and related effects of butadiene.

The genotoxicity of butadiene has been reviewed previously (IARC, 1999).

The frequency of DNA single-strand breaks was increased in NMRI mice exposed to butadiene in vivo.

Decreases in cloning efficiency of T cells and a significant increase in Hprt mutation frequency was observed in female B6C3F₁ mice exposed to 3 ppm [6.63 mg/m³] butadiene for 2 weeks and in rats exposed to 62.5 ppm [138.1 mg/m³] butadiene for 2 weeks; these are the currently reported lowest-observed-effect levels for butadiene that induced Hprt mutations in mice and rats, respectively. The Hprt mutation frequency was also increased in Ephx-1 null and Xpc (DNA repair enzyme) null mice exposed to butadiene for 4 weeks compared with normal mice.

Characterization of Hprt mutations in cDNA and genomic DNA from splenic T-cell mutants was carried out in male B6C3F₁ mice and Fischer 344 rats. The mean Hprt mutant frequency in mice was 9.20 ± 3.25 × 10⁻⁶ in the butadiene-exposed animals compared with 1.48 ± 0.84 × 10⁻⁶ (p < 0.001) in the controls. In rats, the mean Hprt mutant frequency in the exposed animals was 8.08 ± 2.82 × 10⁻⁶ compared with 3.07 ± 0.98 × 10⁻⁶ (p < 0.001) in the controls. In mice, real time PCR and cDNA sequencing showed a statistically significant difference in the overall proportion of mutation types detected in control versus butadiene-exposed mice (p = 0.042), while multiplex PCR of genomic DNA showed that deletion mutations were significantly increased in butadiene-exposed mice (p = 0.031). Analysis of the individual mutation types that occurred in both treated and control mice showed that exposure to butadiene significantly increased the frequencies of each type of base substitution, except for A:T→G:C transitions, and frameshifts and deletions. In male rats, no difference in the overall mutational spectra was detected using cDNA sequencing alone or when combined with multiplex PCR. Statistical analyses of the individual mutation types showed that exposure to butadiene significantly increased base substitution at A:T→T:A transversions in both mice and rats, and single-base insertions, deletion mutations and complex mutations as well as G:C→C:G transversions in mice.

In the bone marrow of B6C3F₁ LacI transgenic mice exposed to butadiene, a significant increase in point mutations was observed at A:T base pairs compared with air-exposed mice. In the spleen of the exposed mice, a significant increase in base substitution mutations (GC→AT transitions and GC→TA transversions at non-CpG sites) was detected, and a significant increase in A:T base pairs occurred, similar to that observed in the bone marrow.

Micronucleus formation was increased in NMRI mice and male and female B6C3F₁/CrBR mice exposed to butadiene. Male B6C3F₁ mice were exposed to butadiene and the bone marrow was harvested 24 h after onset of exposure (Jackson et al., 2000). The frequencies of micronucleated polychromatic erythrocytes were significantly increased (28.2 ± 3.1/100 cells; p < 0.05) compared with controls (9.87 ± 2.1/100 cells) as determined by acridine orange staining. Two other methods to determine micronucleus frequency also showed similar results.

The contribution of CYP2E1 to the genotoxicity of butadiene was investigated by pretreating animals with 1,2-trans-dichloroethylene, a selective CYP2E1 inhibitor, and 1-aminobenzotriazole, an irreversible inhibitor of several CYPs, before exposure to butadiene (Jackson et al, 2000). Pretreatment with 1,2-trans-dichloroethylene significantly lowered the micronucleus frequencies observed (19.8 ± 2.5) but the levels were still elevated over those in unexposed controls (11.5 ± 2.0). Pretreatment with 1-aminobenzotriazole caused the micronucleus frequency to fall to a level similar to that in unexposed animals. The frequency of kinetochore-negative micronuclei was also significantly increased in butadiene-exposed animals (21.3 ± 1.2) compared with controls (5.2 ± 1.7), indicating that butadiene is a clastogen.

(b) Butadiene metabolites

(i) Epoxybutene (see Table 27 for details and references)

Table 27

Genetic and related effects of epoxybutene.

Epoxybutene is mutagenic in Salmonella typhimurium in the presence and absence of a metabolic activation system.

The cII mutant frequency was increased only in Big Blue™ mice but not rat fibroblasts in vitro. The in-vitro mutational spectrum induced by epoxybutene (1 mM for 24 h) in Rat2 LacI cells was compared with a background spectrum. Significant increases in GC→AT transitions at non-CpG sites as well as GC→TA transversions (32% of the epoxybutene-induced mutations) were observed. At A:T base pairs, a significant increase was observed in AT→CG and AT→TA transversions.

Epoxybutene increased micronucleus formation in both Big Blue rat and mouse fibroblasts.

In a human B-lymphoblastoid cell line that does not express active CYP2E1, the mutational spectrum at HPRT induced by epoxybutene showed a significant increase in G:C→A:T and A:T→T:A mutations. The 2S-stereoisomer increased the HPRT mutation frequency in human lymphoblastoid TK6 cells whereas the 2R-isomer caused significant increases at higher doses only.

Mean frequencies of sister chromatid exchange were significantly increased in human whole blood lymphocyte cultures by epoxybutene in both the GSTM1-null (p < 0.001) and GSTM1-positive (p = 0.03) genotypes compared with unexposed controls.

The effect of GSTM1 genotype as well as an adaptive dose of epoxybutene on the induction of sister chromatid exchange was examined. Without an adaptive dose, both genotypes showed a significant increase in sister chromatid exchange over controls. With the adaptive dose, the mean number of sister chromatid exchanges was significantly higher following exposure to epoxybutene in the GSTM1-null group (17.42 ± 2.43; p = 0.01) than in the control group (7.69 ± 1.00) or the GSTM1-positive group (14.07 ± 4.22). The results show an increased sensitivity of GSTM1-null subjects to the induction of sister chromatid exchange by epoxybutene.

Epoxybutene did not induce sister chromatid exchange or chromosomal aberrations in human blood lymphocytes when added at the G₀ stage of the cell cycle. However, from further studies in which excision repair was inhibited, it is probable that epoxybutene induces chromosomal aberrations and DNA damage that is repaired by the excision process in G₀ lymphocytes (Kligerman et al., 1999a,b).

In vivo, a significant increase in mutation frequency was observed at the Hprt locus in splenic T cells from female B6C3F₁ mice exposed to epoxybutene. Exposure to epoxybutene by inhalation resulted in an approximately threefold increase in the frequency of LacI mutants in the lungs of female B6C3F₁ LacI transgenic mice (8.3 ± 3.0 × 10⁻⁵ and 9.9 ± 3.0 × 10⁻⁵) compared with air-exposed controls (3.1 ± 0.7 × 10⁻⁵ and 3.6 ± 0.7 × 10⁻⁵). Significant increases in GC→AT transitions at CpG sites (p = 0.001) were detected. A number of other alterations (insertions, deletions and tandem changes) were also increased in the lungs of exposed mice (10/54, 10%) compared with controls (2/59, 4%). When these alterations were considered separately, only the frequency of deletions was significantly increased (p = 0.005).

(ii) Epoxybutanediol (see also Table 28 for details and references)

Table 28

Genetic and related effects of epoxybutanediol.

Epoxybutanediol has the least mutagenic potency of the butadiene epoxides in traditional mutagenic assays.

Epoxybutanediol increased cII mutant frequency in Big Blue™ mouse but not rat fibroblasts.

Mutational spectra of epoxybutanediol were obtained in the Hprt locus in Chinese hamster ovary-K1 cells. Of the 41 mutants analysed, 25 (61%) were base substitutions and 16 (39%) were deletions. The most common base substitutions were GC→AT and AT→GC transitions. Among the deletions, the majority of the mutants showed single exon loss.

Epoxybutanediol increased micronucleus formation in Big Blue™ rat fibroblasts in vitro. The same effect was seen in Big Blue™ mouse fibroblasts only at higher concentrations.

Epoxybutanediol weakly suppressed the haematopoietic progenitor clonogenic response in human CD34⁺ bone-marrow cells.

(iii) Diepoxybutane (see also Table 29 for details and references)

Table 29

Genetic and related effects of diepoxybutane.

Diepoxybutane has been shown to be the most mutagenic of the butadiene epoxides in traditional mutagenicity assays.

Diepoxybutane is mutagenic in S. typhimurium, Escherichia coli and Sulfolobus acidocaldarius in the presence and absence of a metabolic activation system.

Thirty-nine diepoxybutane mutants were analysed in Chinese hamster ovary-K1 cells. Of these, 24 (62%) were base substitutions and 15 (38%) were deletions. The major base substitutions were GC→AT transitions (11/24) and AT→TA (5/24) and GC→CG (6/24) transversions. Among the deletions, the majority of the mutants showed single exon loss.

At low concentrations of diepoxybutane, increases in the frequency of sister chromatid exchange were observed in Big Blue™ mouse and rat fibroblasts and of micronucleus formation in Rat2 LacI cells in vitro (Erexson & Tindall, 2000b; Recio et al, 2000, 2001).

Increases in the HPRT and thymidine kinase (TK) mutation frequency in human lymphoblastoid TK6 cells were observed with all isomers (or the 2R,3R-, 2S,3S- and meso-stereoisomers) of diepoxybutane. The mutational spectrum of diepoxybutane at HPRT in human TK6 cells after exposure to racemic diepoxybutane was compared with background. Significant increases in A:T→T:A transversions and partial deletions were detected. Diepoxybutane increased the frequencies of sister chromatid exchange and micronucleus formation in human lymphocytes in vitro. Other genetic alterations such as chromosomal breakage and hyperdiploidy were observed as the concentration of diepoxybutane increased.

The effect of various polymorphisms on the frequency of sister chromatid exhange in whole blood lymphocytes from human volunteers after in-vitro exposure to diepoxybutane was examined. A significant difference in sister chromatid exchange was observed between the GSTT1 -negative and GSTT1 -positive individuals (79.28 ± 23.33 and 58.96 ± 17.44, respectively; p < 0.01). Individuals who were heterozygous for the C2 allele in CYP2E1 (C1/C2), which is associated with higher enzyme activity, had higher levels of sister chromatid exchange (70.83 ± 10.85) compared with C1/C1 individuals (56.83 ± 17.64; p < 0.05). The RAD51 polymorphism as well as EH activity had no effect. In individuals with the GSTT1-null genotype, a significant difference was observed in individuals with very low or low expected EH activity (72.13 ± 12.41) and in individuals whose activity was expected to be high (102.73 ± 21.22). Similar results were observed in an earlier study in which GSTT1-null individuals had a sister chromatid exchange frequency of 84.8 ± 20.3 whereas individuals who were GSTT1 -positive had a frequency of 67.9 ± 10.8 (p < 0.001). In this study, no effect was observed in individuals with different GSTM1 genotypes (Schlade-Bartusiak et al, 2000, 2004).

After in-vivo exposure of female B6C3F1 mice and Fischer 344 rats to diepoxybutane, dose-related increases in mutation frequency at the Hprt locus were observed in the splenic T cells of both species. LacI transgenic mice and rats exposed to diepoxybutane showed no or a weak increase in LacI mutation frequency in the spleen and bone marrow.

4.2.3. Mechanism of mutation induction

(a) DNA adducts

Many adducts with epoxybutene, epoxybutanediol and diepoxybutane have been identified in reactions with nucleosides and DNA in vitro (see Table 30 for details and references). The mutagenicity and mutation spectra of several of these adducts have been investigated (see Table 31 for details and references). Many of these adducts can also block replication by many polymerases or can cause misincorporation of proper nucleotides (see Table 32 for details and references). DNA adducts have been identified in humans exposed to butadiene and in animals exposed to butadiene and its metabolites.

Table 30

Reactivity of butadiene metabolites with DNA bases in vitro.

Table 31

Genetic and related effects of the DNA adducts of butadiene.

Table 32

Effect of the butadiene-derived DNA adducts on replication/repair.

(i) Butadiene

Humans

The levels of the DNA adduct, N-1-(2,3,4-trihydroxybutyl)adenine, was determined in 15 male butadiene-exposed (monomer unit) and 11 male control workers from a butadiene monomer production plant in the Czech Republic (Zhao et al., 2000). The median exposure concentration of butadiene for the exposed group was 0.53 mg/m³ [0.24 ppm] whereas that for the control group was 0.013 mg/m³ [0.006 ppm]. Because of interfering background peaks, the N-1-adenine adducts were converted to N⁶-adenine adducts. This adduct was detected in 14 of 15 exposed workers and five of 11 controls. The difference in the levels of adducts between the butadiene-exposed workers (4.5 ± 7.7 adducts/10⁹ nucleotides) and the control workers (0.8 ± 1.2 adducts/10⁹ nucleotides) was significant (p = 0.038). When controls were subdivided into smokers and nonsmokers, the adduct levels were 1.5 ± 1.7 adducts/10⁹ nucleotides for four smokers and 0.3 ± 0.6 adducts/10⁹ nucleotides for seven nonsmokers but the difference was not significant. A significant correlation between the levels of the N-1-(2,3,4-trihydroxybutyl)adenine adduct in lymphocyte DNA and individual exposures to butadiene was found in the exposed group (r = 0.707; p = 0.005), control (r = 0.733; p = 0.01) and both groups combined (r = 0.723; p < 0.001; Zhao et al., 2001). However, no significant correlations were found with other genotoxic effects such as DNA single-strand breaks or micronucleus formation.

Experimental systems

1,4-Bis(guan-7-yl)-2,3-butanediol is the N7-guanine–N7-guanine crosslink formed from the reaction of DNA with diepoxybutane. This compound has been identified in the livers and lungs of C57BL/6 mice that were exposed to 625 ppm [1381 mg/m³] butadiene by inhalation for 7 h per day for 5 days (Goggin et al., 2007). The DNA from livers and lungs contained 3.2 ± 0.4 and 1.8 ± 0.5 adducts/10⁶ guanines from racemic diepoxybutane but no adducts from meso-diepoxybutane were detected.

Male B6C3F₁ mice and Sprague-Dawley rats were exposed for 6 h to 200 ppm [442 mg/m³] [2,3-¹⁴C]butadiene by nose-only inhalation and were killed 48 h after treatment; the livers and lungs were analysed for DNA adducts (Boogaard et al., 2001b). In the livers and lungs of both rats and mice, N7-(2,3,4-trihydroxybut-1-yl)guanine (G4) was the major adduct detected. Mice had 102 ± 4 and 80 ± 4 G4 adducts/10⁸ nucleotides (means ± SE) and rats had 10 ± 2 and 13 ± 0.2 G4 adducts/10⁸ nucleotides in the liver and lung, respectively. Smaller amounts of N7-(2-hydroxy-3-buten-1-yl)guanine (G1) and N7-(1-hydroxy-3-buten-2-yl)guanine (G2) were detected. Mouse liver and lung contained 21 ± 9 and 3.4 ± 1.3 adducts/10⁸ nucleotides, respectively, and rat liver and lung contained 1.9 ± 0.2 and 3.6 ± 1.2 adducts/10⁸ nucleotides, respectively. In rats, no N7-(1,3,4-trihydroxybut-2-yl)guanine (G3) was detected whereas mouse liver and lung contained 25 ± 2 and 4.3 ± 0.1 adducts/10⁸ nucleotides, respectively. A similar profile was also obtained in animals that were killed immediately after cessation of exposure (Boogaard et al., 2004).

In another study, male B6C3F₁ mice and Sprague-Dawley rats were exposed to 20 ppm [44.2 mg/m³] [2,3-¹⁴C]butadiene for either 6 h or for 6 h per day for 5 days by nose-only inhalation, and the livers, lungs and testes were analysed for DNA adducts (Booth et al., 2004b). Following the single 20-ppm exposure, G4 was the major adduct detected in all tissues in rats and mice. Mice had 13.91 ± 3.64, 13.67 ± 0.97 and 6.04 ± 0.8 G4 adducts/10⁸ nucleotides and rats had 5.75 ± 1.32, 3.31 ± 1.74 and 1.39 ± 0.50 G4 adducts/10⁸ nucleotides in liver, lung and testis, respectively. Small amounts of G3 (the exact identity was not determined) were detected consistently in mouse liver and in only one rat liver sample and was not detected in any other tissue. Following the 20-ppm 5-day exposure, G4 was again the main adduct detected and levels in mouse tissues were higher than those in corresponding rat tissues. G1 and G2 were detected in the liver and lung of mice and rats but not in mouse testis; rats had detectable levels in the testis. G3 was detected in mouse tissues and in the liver of rats. The amounts of these other metabolites were much lower than that of G4.

Female Sprague-Dawley rats were exposed to 1000 ppm [2210 mg/m³] butadiene by inhalation for 6 h per day on 5 days per week for 13 weeks. DNA was isolated from the liver and was analysed for the presence of the α-regioisomer of HMVK-derived 1,N²-propanodeoxyguanosine by LC–MS/MS (Powley et al., 2007). No adducts were detected.

Female B6C3F₁ mice and Fischer 344 rats were exposed by inhalation to butadiene for 6 h per day on 5 days per week for 2 weeks (Oe et al., 1999; Blair et al., 2000). The mean daily concentration of butadiene was approximately 1250 ppm [2762.5 mg/m³]. DNA was isolated and analysed for the presence of (±)-G4 and meso-G4. On exposure day 10, mouse liver had 3.9 and 2.2 (±)-G4 and meso-G4 adducts/10⁶ normal bases, respectively. On the same exposure day, rat liver had 1.6 and 0.8 (±)-G4 and meso-G4 adducts/10⁶ normal bases, respectively. The in-vivo half-lives were 4.1 and 5.5 days for (±)-G4 and meso-G4, respectively, in mouse liver DNA. Half-lives for (±)-G4 and meso-G4 in rat liver DNA were 3.6 and 4.0 days, respectively.

DNA adducts were measured in the livers, lungs and kidneys of Fischer 344 rats and B6C3F₁ mice after exposure to 0, 20, 62.5 or 625 ppm [0, 44.2, 138 or 1381 mg/m³] butadiene for 6 h per day or 5 days per week for 4 weeks (Swenberg et al., 2000b). Adducts corresponding to the N7-guanine adducts of epoxybutene (G1 and G2) and the 2,3,4-trihydroxybutane–guanine adduct (G4) that can be formed from either epoxybutanediol or diepoxybutane were analysed. More G4 was detected in mice and rats than G1 and G2. In rats, adduct levels in the different tissues were similar. In general, mouse tissues contained higher levels of DNA adducts.

DNA adducts were measured in the livers, lungs and kidneys of Fischer 344 rats and B6C3F₁ mice after exposure to 0, 20, 62.5 or 625 ppm [0, 44.2, 138 or 1381 mg/m³] butadiene for 6 h per day on 5 days per week for 4 weeks (Koc et al., 1999). Both racemic and meso-G4 and G1 and G2 were analysed by LC–MS/MS. At 625 ppm [1381 mg/m³] butadiene, mouse liver had 31.9 ± 6.5, 32.2 ± 4.2, 3.0 ± 0.1 and 2.4 ± 0.3 racemic G4, meso- G4, G1 and G2 adducts/10⁶guanine bases, respectively. At 625 ppm butadiene, rat liver had 7.7 ± 4.5, 4.2 ± 2.5, 1.2 ± 0.5 and 0.9 ± 0.5 racemic G4, meso- G4, G1 and G2 adducts/10⁶guanine bases, respectively. The number of adducts for both G4 and G1 and G2 was similar for all three tissues at all doses examined in both rats and mice. Mice had significantly higher amounts of G4 adducts than rats in all three tissues after exposure to 625 ppm. This difference was also significant in lung and kidney at 62.5 ppm but not at 20 ppm. Overall, the amounts of G1 and G2 adducts were lower than those of G4 adducts in both species.

In the lungs of mice exposed to 500 ppm [1105 mg/m³] butadiene for 6 h per day for 5 days, N7-guanine DNA adducts arising from epoxybutene and epoxybutanediol were analysed (Koivisto & Peltonen, 2001). All four epoxybutene-derived adducts were detected, most of which arose from S-epoxybutene. For epoxybutanediol, 75% of the total adducts originated from the 2R-diol-3S-epoxybutene isomer and the reaction occurred almost exclusively at the terminal carbon.

Rats were exposed to 300 ppm [663 mg/m³] butadiene for 6 h per day for 5 days and their liver DNA was analysed for the N⁶-(2,3,4-trihydroxy-but-1-yl)adenine adduct (Zhao et al., 1998). The average level of adduct detected in treated rats was 4.5 adducts/10⁹ nucleotides whereas none was detected in control rat liver.

(ii) Butadiene metabolites in experimental systems

Epoxybutene

Male B6C3F₁ mice and Sprague-Dawley rats received a single intraperitoneal injection of 1–50 mg/kg bw [¹⁴C]epoxybutene and were killed 48 h later (Boogaard et al., 2004). DNA was isolated from liver and lung and analysed for the presence of adducts. No adducts were detected in the lungs in either rats or mice. Adduct levels in the liver were below the limit of detection in rats treated with 1 mg/kg bw and in mice treated with 1 and 5 mg/kg bw epoxybutene. Overall, the adduct profiles were similar in rats and mice but rats had much higher levels of adducts than mice. In mice treated with 21 mg/kg bw epoxybutene, the average concentrations of G1 and G2, G3 and G4 were 368, 28 and 50 adducts/10⁸ nucleotides, respectively. In rats treated with 18 mg/kg bw epoxybutene, the concentrations of G1 and G2, G3 and G4 were 857 ± 291, 21 ± 12 and 101 ± 25 adducts/10⁸ nucleotides, respectively.

The livers and lungs of male B6C3F₁ mice and Sprague-Dawley rats that received an intraperitoneal injection of 20 mg/kg bw [4-¹⁴C]epoxybutene and were killed 48 h later were analysed for DNA adducts. In rats, 857 G1 and G2 adducts/10⁸ nucleotides and 101 G4 adducts/10⁸ nucleotides were detected whereas 368 G1 and G2 adducts/10⁸ nucleotides and 50 G4 adducts/10⁸ nucleotides were detected in mice (Boogaard et al., 2001b). No DNA adducts were detected in the lung.

Butanediol

Female B6C3F₁ mice and Fischer 344 rats were exposed by inhalation to 0–36 ppm [0–129.6 mg/m³] butanediol. DNA was isolated from liver and lung and analysed for the presence of G4 (Powley et al., 2005). Both racemic and meso-isomers of this compound can be formed and values were reported as total G4. Mice had significantly greater amounts of adduct than rats in both liver and lung at 6 and 18 ppm [21.6 and 64.8 mg/m³] but there was little difference between the tissue levels. At 6 ppm, mice had 6030 ± 1740 and 5570 ± 540 fmol adducts/mg DNA whereas rats had 2560 ± 180 and 2320 ± 640 fmol adducts/mg DNA in the liver and lung, respectively. Similarily shaped dose–response curves were observed for G4 and the Hprt mutant frequency in splenic T cells in rodents exposed to butanediol.

However, when female Fischer 344 or Sprague-Dawley rats were exposed to higher concentrations of butanediol (36 ppm or 1000 ppm, respectively, for 6 h per day on 5 days per week for 4 weeks), no adducts were detectable in the DNA of the liver of these animals (Powley et al., 2007).

Diepoxybutane

Diepoxybutane induces the formation of DNA–protein cross-links with the DNA repair protein, O⁶-alkylguanine–DNA alkyltransferase (AGT) (Loeber et al., 2006). The product of initial DNA alkylation by diepoxybutane, N7-(2′-hydroxy-3′,4′-epoxybut-1′-yl)–deoxyguanosine, was incubated with recombinant human AGT and analysed by HPLC–electrospray ionization MS. Analysis of the whole protein showed the presence of a monoalkylated protein and a protein that contained two butanediol cross-links. Peptide mapping revealed that the DNA–AGT cross-link involved the sulfhydryls of Cys¹⁴⁵ or Cys¹⁵⁰ within the human AGT active site and the N7 position of guanine in duplex DNA. No cross-linking was detected with Cys⁵, Cys²⁴ or Cys⁶². The resulting structure was 1-(S-cysteinyl)-4-(guan-7-yl)-2,3-butanediol.

The effect of the stereochemistry of the isomers of diepoxybutane and their abilities to form cross-links with calf thymus DNA was investigated (Park et al., 2005). Comparable amounts of total 1,4-bis-(guan-7-yl)-2,3-butanediol (bis-N7-guanine–butadiene) crosslinks and G4 adducts were observed. However, the types of cross-link (either interstrand or intrastrand) varied depending on the stereoisomer used. S,S-Diepoxybutane produced the highest amount of 1,3 interstrand cross-links (96%) followed closely by racemic diepoxybutane (90%). meso-Diepoxybutane produced almost equal amounts of 1,3 interstrand (49%) and 1,2 intrastrand cross-links (51%). R,R-Diepoxybutane produced 19% 1,2 intrastrand cross-links and 68% 1,3 interstrand cross-links but also produced a large quantity of 1,2 interstrand cross-links (13%) that were not detected with the other stereoisomers.

N⁶-(2-Hydroxy-3,4-epoxybut-1-yl)adenine can potentially be produced from the reaction of diepoxybutane with DNA. This compound as well as its corresponding DNA oligomer have been synthesized (Antsypovich et al., 2007). Yields of the compounds were lower than expected and both readily cyclized to an unidentified exocyclic diepoxybutane–deoxyadenosine side-product. Formation of cross-links by this compound was minimal. The half-life of N⁶-(2-hydroxy-3,4-epoxybut-1-yl)adenine in single-stranded DNA was < 2 h at physiological conditions.

Calf thymus DNA incubated with a 40-fold molar excess of epoxybutene resulted in the detection of equimolar amounts of the two N7-guanine adducts of epoxybutene (G1 and G2) (Blair et al, 2000); when diepoxybutane was used, the (±)-G4 adduct was detected as the major product. When human TK6 cells were exposed to 400 µM epoxybutene for 24 h, the concentration of G1 and G2 adducts was 4.3 ± 0.9 and 4.1 ± 1.0 adducts/10⁶normal cells, respectively. Urine samples from Fischer 344 rats and B6C3F1 mice exposed to 1250 ppm [3453 mg/m³] butadiene were analysed for the presence of N7-guanine adducts. For all 3 days on which the adducts were analysed, rats excreted significantly more G1 and G2 and (±)-G4; G1 was the major adduct excreted. In mice, a small amount of meso-G4 was detected but none was found in rat urine.

DNA adducts were analysed in MCF-7 cells after a 6-h exposure to epoxybutanediol and diepoxybutane at concentrations of 100-1000 µmol/mg DNA (Koivisto et al, 1999). At all concentrations tested, more diepoxybutane– N7-guanine adducts were detected. In the lungs of mice exposed to 50–1300 ppm [110.5–2873 mg/m³] butadiene for 6 h per day for 5 days, large amounts of epoxybutanediol– N7-guanine (G4) adducts were detected. A small peak for the adduct between diepoxybutane and guanine at N7 was detected; however, this peak also elutes closely with an epoxybutene-derived adduct.

4.2.4. Alterations in oncogenes and suppressor genes in tumours

A specific codon 13 mutation in K-ras has been described previously (see IARC, 1999).

Lymphomas induced in B6C3F₁ mice by exposure to butadiene were analysed for gross structural alterations and point mutations in several proto-oncogenes that are implicated in the ras, p53 or pRb pathways (Zhuang & Söderkvist, 2000). Using southern blotting, no structural alterations or amplifications were detected in Raf1, Mdm2, c-Myc, Cdc25a or Cdc25b. Ten tumours exhibited four identical silent base substitutions (GAC522AAC, GTG531GTC, TCG533TCC, GCT543ACT) and allelotypic analysis showed loss of heterozygosity of the Raf1 locus in six of 31 butadiene-induced lymphomas. No changes were detected in lymphomas induced by long-term inhalation of 20–625 ppm [44.2–1381 mg/m³] butadiene in B6C3F₁ mice that were analysed for genetic alterations in Rb1, Ccnd1 and Cdk4 genes. These results suggest that, if the inactivation of other tumour-suppressor genes may be involved in the development of a subset of butadiene-induced lymphomas, the genetic alterations in the above proto-oncogenes do not play an important role in the development of these tumours.

Studies using the supF gene and the three diepoxybutane isomers showed that S,S-diepoxybutane was the most potent mutagen followed by R,R-diepoxybutane and meso-diepoxybutane (Kim et al., 2007). The major form of mutation was A:T→T:A transversion following treatment with all three stereoisomers. However, S,S-diepoxybutane induced larger numbers of G:C→A:T transitions while R,R-diepoxybutane resulted in a higher frequency of G:C→T:A transversions.

Rarely is the nervous system a target in chemical carcinogenesis but, in B6C3F₁ mice exposed to 625 ppm [1381 mg/m³] butadiene, six malignant gliomas and two neuroblastomas were observed. Only one tumour has been reported in more than 500 historical control mice. Morphologically, the characteristics of the malignant gliomas and neuroblastomas were consistent with those reported for humans. Tumours were also evaluated for genetic alterations in p53, K- and H-ras genes (Kim et al., 2005). One neuroblastoma had a mutation in codon 61 of the H-ras gene. Missense mutations in p53 exons (exons 5–8) were detected in two neuroblastomas and three of six malignant gliomas and consisted mostly of G→A transitions (5/6). Three of five malignant gliomas and both neuroblastomas showed nuclear accumulation of p53 protein. Loss of heterozygosity at the p53 gene locus was also observed in four of five malignant gliomas and both neuroblastomas. All of these tumours displayed loss of the C57 (B) allele at the Ink4a/Arf gene locus, which codes for the p16^Ink4a that may play a key role in the development of mouse brain tumours.

Male and female B6C3F₁ mice were exposed to a total of 8100 ppm [17 901 mg/m³]– weeks and 16 200 ppm [35 802 mg/m³]–weeks butadiene (Ton et al., 2007). Of 51 lung tumours, 34 had a GGC→CGC transversion mutation in the codon 13 of the K-ras gene. A loss of heterozygosity on chromosome 6 was observed in 14 of 51 of the lung tumours and all were in the region of K-ras. There is mounting evidence that K-ras, which is a known oncogene, functions as a tumour-suppressor gene and its loss may play a major role in mouse lung carcinogenesis.

Point mutations in ras genes were analysed in forestomach tumours from male and female B6C3F₁ mice that had been exposed to 6.25–625 ppm [13.8–1381 mg/m³] butadiene by inhalation for 6 h per day on 5 days per week for 1–2 years (Sills et al., 2001). Among the butadiene-induced tumours, 20 of 24 contained ras gene mutations compared with four of 11 spontaneous forestomach neoplasms. In butadiene-induced forestomach tumours, the most common transversions were GGC→CGC in codon 13 in K-ras, and CAA→CTA in codon 61 in H-ras.

In male and female B6C3F₁ mice exposed to 6.25–625 ppm [13.8–1381 mg/m³] butadiene by inhalation for 6 h per day on 5 days per week for 2 years, an increase in the incidence of cardiac haemangiosarcomas was observed. Eleven of these haemangio-sarcomas were analysed for alterations in the p53 gene and ras proto-oncogenes (Hong et al., 2000). Most of the butadiene-induced haemangiosarcomas (9/11) had K-ras mutations and all nine had G→C transversions in codon 13. Of the nine haemangiosarcomas with the K-ras mutations, five also had an H-ras codon 61 CGA mutation. Mutations in exons 5–8 of the p53 gene were identified in five of 11 haemangiosarcomas.

Butadiene-induced mammary tumours (17) were collected from female B6C3F₁ mice that were exposed to 6.25–625 ppm [13.8–1381 mg/m³] butadiene by inhalation for up to 2 years (Zhuang et al., 2002). Genetic alterations in the p53 gene were found in seven of 17 tumours. All of these tumours also showed loss of the wild-type p53 allele. Missense mutations in codons 12, 13 or 61 of the H-ras gene were found in nine of 17 tumours. Seven of these nine H-ras mutations were G→C transversions in the first base of codon 13. Missense mutations in the b-catenin gene were identified in three of 17 tumours. No mutations were identified in the Apc or Axin genes.