Reaction mass of butane and butene

Carcinogenicity

Description of key information

Reaction mass of butane and butene containing > 0.1 % butadiene: carcinogenic
Reaction mass of butane and butene containing < 0.1 % butadiene: not carcinogenic

Justification for classification or non-classification

1,3-butadiene is classiefied in Annex VI of the CLP as Carc. Cat. 1A (formerly R45) and Muta. Cat. 1B (formerly R46). According to the CLP, "Reaction mass of butane and butene" has to be classified as Carc. Cat. 1A (formerly R45) and Muta. Cat. 1B (formerly R46) if the content of 1,3 -butadiene is equal to or above 0.1 %. Testing for genetic toxicity and carcinohenicity is therefore not required in accordance with Column 2 of REACH Annex VII and Annex X.

If the content of 1,3 -butadiene is below 0.1 %, no classification and labeling is required as the available data for the other components (butene, butane, 2 -methylpropene and isobutane) shows that they are neither mutagenic nor that a carcinogenic effect has to be expected.

Additional information

There is no data available on carcinogenicity of "Reaction mass of butane and butene". Below we present an assessment of the data for each of the single components Butene, 2 -methylpropene, butane, isobutane and 1,3 -butadiene which are present in "Reaction mass of butane and butene".

Butene/2-methylpropene:

 

Members of the butenes category are flammable gases at room temperature and therefore significant exposure via the dermal or oral routes is unlikely. Inhalation exposure is the most appropriate route for carcinogenicity testing.

 

Carcinogenicity studies via inhalation exposure are available for 2-methylpropene (also called isobutylene). F344 Rats and B6C3F1 mice were exposed by inhalation at 0, 500, 2,000 or 8,000 ppm (1147, 4589, 18,359 mg/m³), for 105 weeks (NTP 1998). In rats, treatment-related non-neoplastic findings were limited to increased kidney weights, hyaline degeneration of the olfactory epithelium and hypertrophy of goblet cells lining the nasopharyngeal duct in rats. In mice, treatment-related non-neoplastic findings were limited to increased hyaline degeneration of the olfactory and respiratory epithelium.These findings were generally considered to be of low toxicological significance and are discussed in detail in Section 7.5 (Repeated dose toxicity). The major urinary metabolite of 2-methylpropene (2-hydroxyisobutyric acid: HIBA) was measured in the urine of both species as an indicator of isobutene exposure, these results are discussed in Section 7.1 Toxicokinetics.

 

There were no treatment-related neoplasms in female rats or male and female mice. The NTP concluded that there was no evidence of carcinogenic activity of isobutene in female rats or male and female mice exposed to 500, 2,000, or 8,000 ppm. A NOAEC of 8000 ppm (18,359 mg/m3) for carcinogenicity in female rats, male mice and female mice was established based on these results. 

 

The incidence of thyroid gland follicular cell carcinoma in male rats exposed to 8000 ppm was increased compared to the chamber control group (1/48, 0/48, 0/48, 5/50 at 0, 500, 2000 and 8000 ppm respectively) and exceeded the historical control range. The NTP therefore concluded that there was some evidence of carcinogenic activity of 2-methylpropene in male rats and established a NOAEC of 2000 ppm (4589 mg/m3) for carcinogenicity in male rats based on the thyroid follicular cell carcinomas. The relevance of the thyroid follicular tumours for human cancer risk is questionable as the tumours occurred in male rats only and not in either sex in mice. There were no precursor lesions, no dose-response relationship, the tumours were singular and unilateral, did not form metastases and there was no increase in liver weight indicating a secondary mechanism. In addition, the thyroid was not a target organ in repeat dose studies in rats. Although the 10% incidence of tumours was outside the historical control at the time, reported to be 0-4% (NTP 1998, Haseman et al 1998), a carcinogenicity study on propylene (NTP 1985) had a 7% incidence of thyroid follicular carcinomas in the control group, further diminishing the significance of the incidence with 2-methylpropene and suggesting that the tumor findings may have been spurious. IARC (1999) published guidance noting that no non-radioactive chemical exposure is known to cause thyroid follicular carcinomas in humans. It concluded that agents causing thyroid neoplasms in rodents by hormonal imbalance must be non-genotoxic. Although studies investigating hormonal imbalance have not been conducted, 2-methylpropene is not genotoxic, indicating if 2-methylpropene did cause an increase in thyroid tumors in male rats, that a non-genotoxic mechanism, likely to have a threshold, is involved and the relevance of tumours of this type to human health is low.

  

There is no data on the carcinogenicity of the butenes in humans. Toxicokinetic data on members of the butenes category (see Toxicokinetics Section), indicate that humans have the lowest capacity for oxidative metabolism and the highest for detoxification pathways.

In conclusion, as no tumours were observed in female rats or male and female mice and the thyroid tumours observed only in male rats were either spurious or of questionable relevance to humans. As all members of the butenes category are not genotoxic, members of the butenes category have a low potential for human carcinogenicity.

 

 

Additional References:

Haseman JK, Hailey JR, Morris and RW (1998). Spontaneous neoplasm incidences in Fischer 344 rats and B6C3F1 mice in two-year carcinogenicity studies: a National Toxicology Program update. Toxicol Pathol, 26, 428-441

 

NTP (1985). NTP Toxicology and Carcinogenesis Studies of Propylene (CAS No. 115-07-1) in F344/N Rats and B6C3F1 Mice (Inhalation Studies).Natl Toxicol Program Tech Rep Ser, 272, :1-146

 

IARC Sci Publ. (1999). Agents that induce epithelial neoplasms of the urinary bladder, renal cortex and thyroid follicular lining in experimental animals and humans: summary of data from IARC monographs volumes 1-69. ;Ed Wilbourn JD, Partensky C, Rice JM. 147,191-209.

Butane/isobutene:

There are no carcinogenicity studies available for any of the C1 - C4 alkane gases which comprise the Petroleum Gases category. However, weight of evidence from subchronic tests (up to 90 days), a consideration of their simple chemical structures, which have no reactive groups and carry no alerts for likely genotoxic carcinogenic activity from established Structure Activity Relationship analysis ( Tennant RW and Ashby J (1991). Classification according to chemical structure, mutagenicity to Salmonella and level of carcinogenicity of a further 39 chemicals by the US National Toxicology Program.  Mutat Res 257 (3) 209-227), together with the conclusion that C1-C4 alkanes are not genotoxic, provide a strong case for concluding that none will show any significant carcinogenic activity. Taking these data into account, together with the general lack of toxicity across other endpoints, it is considered that there is no justification for conducting further animal carcinogenicity studies. The above reasoning leads to the conclusion that Petroleum Gases category can be considered to have low concern for human carcinogenicity.

1,3-butadiene:

 

The carcinogenicity of 1,3-butadiene has been extensively reviewed, including an EU Risk Assessment Report (2002), ECETOC (1997), SCOEL (2007), US EPA (2002) and TCEQ (2008). The non-human data in this endpoint summary is based on the EU RAR (2002) as there have been no new animal data since 2002. The human information has been updated as new information has become available since 2002. The carcinogenicity of 1,3-butadiene in humans is regarded to be the most important health effect of this chemical and the rationale for the development of the worker and population DMELs is described.

 

Non-human information

 

The carcinogenicity of 1,3-butadiene has been studied in rats and mice.

 

In the rat an inhalation study was conducted on behalf of the International Institute of Synthetic Rubber Producers (IISRP) (Owen et al 1987). Groups of male and female rats were exposed to 1,3-butadiene at 1000 or 8000 ppm (2212 or 17701 mg/m³) for 6 hr/day, 5 days/week for 2 years. There were increases in the incidences of pancreatic exocrine adenoma (high dose, male); uterine sarcoma (both doses, female); Zymbal gland carcinoma (high dose, female); mammary tumours (both doses, female); thyroid follicular cell tumours (both doses female) and testis Leydig-cell tumours (high dose). These data suggest that 1,3-butadiene is a weak carcinogen to the rat under the conditions of exposure used in this study. The increased incidence of mainly benign tumours, which occur spontaneously in the rat, suggests that 1,3-butadiene may act by a non-genotoxic mechanism, rather than by a direct effect of reactive metabolites.

 

The US National Toxicology Program has conducted two carcinogenicity studies in mice. In the first (NTP, 1984), male and female B6C3F1mice were exposed to 1,3-butadiene by inhalation at 625 or 1250 ppm (1382 or 2765 mg/m³), 6 hrs day, 5 days per week for 61 weeks. The study was scheduled for 2 years but was stopped earlier because of high mortality in both treated groups. There was clear evidence of multiple organ carcinogenicity for 1,3-butadiene in both sexes, as shown by increased incidences and early induction of haemangiosarcomas of the heart, malignant lymphomas, alveolar/bronchiolar adenomas and carcinomas, and papillomas of the stomach in males and females; and of acinar cell carcinomas of the mammary gland, granulosa cell tumours of the ovary, and hepatocellular adenomas and carcinomas in females. This study demonstrated that 1,3-butadiene is a potent carcinogen in mice causing multi-organ tumours that develop after only 1 year.

 

The second study extended the dose range of the first and also included a “Stop-Exposure” study where mice were exposed for a period then left untreated (NTP 1993). Survival in treated groups was reduced in both standard and “Stop-Exposure” studies due to the presence of malignant neoplasms. The standard and “Stop-Exposure” studies confirmed the clear evidence of carcinogenicity of 1,3-butadiene in both sexes. In the standard study, male and female B6C3F1mice were exposed to 1,3-butadiene by inhalation at 6.25, 20, 62.5, 200 or 625 ppm (13, 44, 138, 442 or 1382 mg/m³), 6 hrs day, 5 days per week for up to 2 years. Tumours arose at all exposure levels. In males there were increased incidences of neoplasms in the haematopoietic system, heart, lung, forestomach, liver, harderian gland, preputial gland, brain and kidney. In females there were increased incidences of neoplasms in the haematopoietic system, heart, lung, forestomach, liver, harderian gland, ovary and mammary gland. Low incidences of intestinal carcinomas in male mice, Zymbal's gland carcinomas in male and female mice, and renal tubule adenomas and skin sarcomas in female mice may also have been related to administration of 1,3-butadiene. In the “Stop-Exposure” study male B6C3F1 mice were exposed to 1,3-butadiene by inhalation at 200 ppm (443 mg/m³) for 40 weeks, 312 ppm (690 mg/m³) for 52 weeks, 625 ppm (1383 mg/m³) for 13 weeks, or 625 ppm (1383 mg/The carcinogenicity of 1,3-butadiene has been extensively reviewed, including an EU Risk Assessment Report (2002), ECETOC (1997), SCOEL (2007), US EPA (2002) and TCEQ (2008). The non-human data in this endpoint summary is based on the EU RAR (2002) as there have been no new animal data since 2002. The human information has been updated as new information has become available since 2002. The carcinogenicity of 1,3-butadiene in humans is regarded to be the most important health effect of this chemical and the rationale for the development of the worker and population DMELs is described. Non-human information The carcinogenicity of 1,3-butadiene has been studied in rats and mice. In the rat an inhalation study was conducted on behalf of the International Institute of Synthetic Rubber Producers (IISRP) (Owen et al 1987). Groups of male and female rats were exposed to 1,3-butadiene at 1000 or 8000 ppm (2212 or 17701 mg/m³) for 6 hr/day, 5 days/week for 2 years. There were increases in the incidences of pancreatic exocrine adenoma (high dose, male); uterine sarcoma (both doses, female); Zymbal gland carcinoma (high dose, female); mammary tumours (both doses, female); thyroid follicular cell tumours (both doses female) and testis Leydig-cell tumours (high dose). These data suggest that 1,3-butadiene is a weak carcinogen to the rat under the conditions of exposure used in this study. The increased incidence of mainly benign tumours, which occur spontaneously in the rat, suggests that 1,3-butadiene may act by a non-genotoxic mechanism, rather than by a direct effect of reactive metabolites. The US National Toxicology Program has conducted two carcinogenicity studies in mice. In the first (NTP, 1984), male and female B6C3F1mice were exposed to 1,3-butadiene by inhalation at 625 or 1250 ppm (1382 or 2765 mg/m³), 6 hrs day, 5 days per week for 61 weeks. The study was scheduled for 2 years but was stopped earlier because of high mortality in both treated groups. There was clear evidence of multiple organ carcinogenicity for 1,3-butadiene in both sexes, as shown by increased incidences and early induction of haemangiosarcomas of the heart, malignant lymphomas, alveolar/bronchiolar adenomas and carcinomas, and papillomas of the stomach in males and females; and of acinar cell carcinomas of the mammary gland, granulosa cell tumours of the ovary, and hepatocellular adenomas and carcinomas in females. This study demonstrated that 1,3-butadiene is a potent carcinogen in mice causing multi-organ tumours that develop after only 1 year. The second study extended the dose range of the first and also included a “Stop-Exposure” study where mice were exposed for a period then left untreated (NTP 1993). Survival in treated groups was reduced in both standard and “Stop-Exposure” studies due to the presence of malignant neoplasms. The standard and “Stop-Exposure” studies confirmed the clear evidence of carcinogenicity of 1,3-butadiene in both sexes. In the standard study, male and female B6C3F1mice were exposed to 1,3-butadiene by inhalation at 6.25, 20, 62.5, 200 or 625 ppm (13, 44, 138, 442 or 1382 mg/m³), 6 hrs day, 5 days per week for up to 2 years. Tumours arose at all exposure levels. In males there were increased incidences of neoplasms in the haematopoietic system, heart, lung, forestomach, liver, harderian gland, preputial gland, brain and kidney. In females there were increased incidences of neoplasms in the haematopoietic system, heart, lung, forestomach, liver, harderian gland, ovary and mammary gland. Low incidences of intestinal carcinomas in male mice, Zymbal's gland carcinomas in male and female mice, and renal tubule adenomas and skin sarcomas in female mice may also have been related to administration of 1,3-butadiene. In the “Stop-Exposure” study male B6C3F1 mice were exposed to 1,3-butadiene by inhalation at 200 ppm (443 mg/m³) for 40 weeks, 312 ppm (690 mg/m³) for 52 weeks, 625 ppm (1383 mg/m³) for 13 weeks, or 625 ppm (1383 mg/m³) for 26 weeks. After exposure the mice were then left untreated for the remainder of the 2-year study. Tumours at multiple sites were observed at all dose levels with the first tumours appearing after only 13 weeks of exposure to 650 ppm. These NTP studies (standard and "Stop Exposure") also show that 1,3-butadiene causes multi-site carcinogenicity in mice. Tumours arose at all exposure levels. These data indicate that 1,3-butadiene is a genotoxic carcinogen and the risk of carcinogenicity in mice is high even at low exposure levels (NTP 1984, 1993). A final study in mice was conducted by Bucher (1993). Male and female B6C3F1 mice were exposed to 1,3-butadiene for a single 2-hour period to concentrations of 0, 1000, 5000 or 10,000 ppm (2212, 11063 or 22126 mg/m³). The mice were then held for 2 years without treatment. There were no effects on survival at 2 years, no effects on bodyweight and no increased incidences of neoplastic or non-neoplastic lesions attributed to exposure to 1,3-butadiene. Although this study did not identify any carcinogenic effect associated with an acute exposure to 1,3-butadiene the studies with multiple exposures are more relevant for hazard assessment. In conclusion, there is a marked species difference in the carcinogenicity of 1,3-butadiene in experimental animals. The difference in response is consistent with the in vivo genotoxicity of 1,3-butadiene in mice but not rats. In the mouse, 1,3-butadiene is a potent multi-organ carcinogen. Tumours develop after short durations of exposure, at low exposure concentrations and the carcinogenic response includes rare types of tumours. In the rat fewer tumour types, mostly benign develop at exposure concentrations of 100 to1000-times higher than in the mouse. The tumour response in rats suggests that non-genotoxic, possibly hormonal mechanisms influence the carcinogenicity in this species.   Human information The European Union Risk Assesssment Report (EC, 2002) concluded that there is clear evidence from a study in styrene-butadiene rubber (SBR) workers (Delzell et al., 1995, 1996; Macaluso et al., 1996), that occupational exposure to 1,3-butadiene is associated with an excess of leukaemia. IARC (2008) concluded that there is sufficient evidence in humans for the carcinogenicity of 1,3-butadiene and noted that this conclusion was based primarily on the evidence for a significant exposure–response relationship between exposure to butadiene and mortality from leukaemia in an update of the SBR workers study (Sathiakumar et al., 2005; Graff et al., 2005: Delzell et al., 2006). More recently IARC have added a statement that “1,3-Butadiene causes cancer of thehaematolymphaticorgans” to their evaluation of the evidence from human studies (Baan et al., 2009).  The SBR workers study provides good quality information on the association between exposure to 1,3-butadiene and haematolymphatic cancer (HLC) for a large group of over 16,000 workers with a long period of follow up, and the US EPA (2002) concluded that it provided the best published set of data to evaluate human cancer risk from 1,3-butadiene exposure, although the EU RAR (2002) stated that “overall these modelled data cannot be viewed as of sufficient reliability on which to base an estimate of the dose response relationship for the carcinogenic effect”. However, the most recent study update incorporates improved exposure estimates for 1,3-butadiene and estimates for potential confounders, styrene and dimethyldithiocarbamate (DMDTC), were calculated (Macaluso et al. 2004). The improved exposure estimates were validated by Sathiakumar et al. (2007). IARC (2008) discussed evidence for an association between butadiene and non-Hodgkin lymphoma which derives from the studies of workers in the monomer industry, and noted that they were unable to determine the strength of the evidence for particular histological subtypes of lymphatic and haematopoietic neoplasms because of changes in coding and diagnostic practices. The study by Divine and Hartman (2001) provides the most reliable information about the association between NHL and 1,3-butadiene exposure in monomer workers, but survival analyses showedno increase in risk with increasing cumulative1,3-butadieneexposurefor NHL and all HLC. Sathiakumar et al (2005) reported that they did not find any clear relation between employment in the SBR industry and other haematolymphatic cancers (besides leukaemia) and reported no excess of deaths from NHL and multiple myeloma.Graff et al.(2005) examined exposure-response trends for the same workers and reported a positive association between 1,3 -butadiene and leukaemia that was not explained by exposure to other agents examined, but they did not report similar associations for NHL and multiple myeloma. Estimates of excess leukaemia risk have been derived using the SBR workers study by various groups (SCOEL, 2007; Sielken et al., 2007, 2008; TCEQ, 2008), but no estimates have been derived for HLC. Given the lack of association with other types of HLC, it seems unlikely that modelling the association between HLC and 1,3-butadiene in the SBR workers would provide better estimates of excess risk than modelling the association between leukaemia and 1,3-butadiene. Cheng et al (2007) used Cox regression procedures to examine the exposure–response relationship between several time-dependent 1,3-butadiene exposure indices and lymphoid neoplasms and myeloid neoplasms in addition to leukaemia. They concluded that evidence of an association between 1,3-butadiene and all lymphoid neoplasms or all myeloid neoplasms is less persuasive than that for all leukaemias. Graff et al. (2005) and Sielken et al. (2008) reported no association between acute myelogenous or monocytic leukaemia (AML) and 1,3-butadiene exposure, indicating that non-AML leukaemia may be a better endpoint than all leukaemia. However, there is limited information available to estimate the number of excess leukaemias of this type and quantitative risk assessment in the SBR workers cohort is based on models using all leukaemia as the endpoint. SCOEL (2007) agreed that 1,3-butadiene should be treated as a possible human carcinogen, operating via a genotoxic mechanism. Excess risk entailed in exposure during a working life to various concentrations of 1,3-butadiene was calculated using a “step” approach (Zocchetti et al, 2004) for 23 sets of model parameters taken from Delzell et al. (2001). These dose response analyses for the SBR cohort incorporated the more refined exposure estimates of Macaluso et al. (2004), but not the additional 7 years of follow up of the latest study update. SCOEL estimated that occupational exposure to 1 ppm of BD for a working life (40 years between the ages of 25 and 65), will cause from 0.0 to 107.8 extra leukaemia deaths per 10⁴workers between the ages 25-85 years. However, 12 of the 23 SCOEL estimates are based on models which ignore exposure to BD at concentrations either below 100 ppm or above 100 ppm, and are not appropriate for risk assessment. With these 12 models excluded, the estimates for 1,3-butadiene exposure of 1 ppm range up to 15.3 per 10⁴excess deaths. However, the major weakness of the step model is that excess death estimates for low exposures are based on a single relative risk (RR) estimate which may have considerable variability. For example, the model giving the highest valid estimate of excess deaths at 1 ppm also gives the same estimate for all 1,3-butadiene exposures < 2.1 ppm, and these are based on a RR of 1.3 with 95% CI (0.4-4.3). In addition, the SCOEL approach doesn’t give a true range of estimates, especially for low exposures, as they are based on RR from different analyses which are highly correlated. Sielken et al.(2008) used the SCOEL assumptions about the relevant exposure window and the same life table assumptions about mortality rates and survival probabilities as SCOEL (SCOEL, 2007) to calculate estimates of occupational risk. This report and an earlier report (Sielken et al.,2007) used Poisson and Cox regression models to model the association between 1,3-butadiene exposure and all leukaemias and leukaemia subtypes. Their models included terms for both cumulative 1,3-butadiene exposure and the cumulative number of exposures to 1,3-butadiene concentrations > 100 ppm (the number of High Intensity Tasks [HITs]). Their results show that cumulative BD HITS is an important predictor of risk with an effect that is independent of cumulative 1,3-butadiene exposure. The EU RAR (2002) had earlier noted that there was some indication that exposures accrued by exposure to 1,3-butadiene peaks may be important in the development of leukaemia, but there was insufficient data to clarify this. It can also be deduced from Graff et al (2005) that there were no leukaemia deaths among 34,152 person years of follow up from SBR workers exposed to 1,3-butadiene but not HITs. Sielken et al. (2008) noted that they considered Cox proportional hazards modelling to be more scientifically appropriate than Poisson regression modelling and Cox regression results are given more weight. They reported that all leukaemia and chronic myeloid leukaemia (CML) were associated with cumulative BD HITs, but not cumulative 1,3-butadiene exposure. Chronic lymphocytic leukaemia (CLL) was associated with cumulative 1,3-butadiene exposure, but not cumulative 1,3-butadiene HITs.Acute myelogenous or monocytic leukaemia (AML) was not associated with either cumulative 1,3-butadiene exposure or cumulative 1,3-butadiene HITs. For 1 ppm exposure and their preferred Cox regression model for leukaemia which adjusted for 1,3-butadiene HITs, Sielken et al (2008) estimated 0.33 extra leukaemia deaths per 10⁴workers. The corresponding estimate from a Poisson regression model that adjusted for 1,3-butadiene HITs was 0.53 extra leukaemia deaths per 10⁴workers. For CLL, the only endpoint significantly associated with cumulative 1,3-butadiene exposure before and after adjustment for 1,3-butadiene HITs, Sielken et al. (2008) derived an estimate of 0.16 extra CLL deaths per 10⁴workers. Cheng et al. (2007) also modelled the leukaemia data from the SBR workers studies using a range of Cox regression models, and further analyses are included in the TCEQ (2008) report. TCEQ (2008) used life table methods to estimate general population lifetime risk estimates using the model parameters derived from these Cox regression models. These risk estimates are not relevant to an occupational exposure scenario, and the regression coefficients of the models are not directly comparable with those fitted by Sielken et al. (2008) which incorporated the SCOEL assumptions about the relevant exposure window and excluded exposures that occurred 40 or more years ago. Nevertheless, the magnitude of a regression coefficient relative to the coefficient of the Cox regression model fitted by Sielken et al. (2008) for which occupational risk estimates are available (continuous cumulative 1,3-butadiene exposure with age and number of 1,3-butadiene HITs > 100 ppm as covariates), can be used to give an approximate estimate of the number of excess leukaemias. Cheng et al. (2007) noted that the slope of the exposure–response relationship was irregular especially in the upper part of the 1,3-butadiene exposure range. In order to reduce the impact of data in the upper part of the cumulative 1,3-butadiene exposure range, Cheng et al. (2007) fitted Cox regression models using either continuous cumulative BD exposure restricted to the lower 95% of exposure range, or the mean scored deciles of cumulative 1,3-butadiene exposure. However, Cheng et al (2007) noted that they preferred the estimate of the exposure–response trend that is based on the continuous, untransformed form of the 1,3-butadiene variables and the full range of exposure data, but noted the high potential for distortion of the exposure–response relationship as a result of exposure misclassification. Table 1 shows Cox regression coefficients for models in which the hazard function was either a log linear function of continuous cumulative 1,3-butadiene exposure or a log linear function of the mean scored deciles of cumulative BD exposure. In addition, the table also shows Cox regression coefficients for the continuous model using the exposure metric that excluded exposure that occurred more than 40 years ago or excluded the 5% of workers with the highest cumulative 1,3-butadiene exposures. Regression coefficients are also shown for models which were fitted with and without BD HITs as covariate. There is little difference between the continuous and categorical regression models that adjust for 1,3-butadiene HITs and do not exclude exposure that occurred more than 40 years ago. All of the regression coefficients were close to that of the preferred model of Sielken et al. (2008) except for that of the mean scored deciles model that did not adjust for 1,3-butadiene HITs. The coefficient of the Poisson regression model derived by Sielken et al. (2008) that did not adjust for 1,3-butadiene HITs (1.76 x 10^-3) was also much higher than the regression model that did adjust for BD HITs (3.42 x 10^-4), and is also based on mean scored deciles of cumulative BD exposure.   Table 1 Values of Maximum Likelihood Estimate ofβ, standard error (SE) and size of coefficient relative to that of the Sielken et al. (2008) Cox regression model with adjustment for1,3-butadieneHITsModelCovariatesAgeAge & Number of 1,3-butadiene HITs > 100 ppmSourceβ(MLE)±SESourceβ(MLE)±SECox log-linear ppm-years continuous, excluding exposure that occurred > 40 years agoSielkenet al.(2008)m³) for 26 weeks. After exposure the mice were then left untreated for the remainder of the 2-year study. Tumours at multiple sites were observed at all dose levels with the first tumours appearing after only 13 weeks of exposure to 650 ppm.

 

These NTP studies (standard and "Stop Exposure") also show that 1,3-butadiene causes multi-site carcinogenicity in mice. Tumours arose at all exposure levels. These data indicate that 1,3-butadiene is a genotoxic carcinogen and the risk of carcinogenicity in mice is high even at low exposure levels (NTP 1984, 1993).

 

A final study in mice was conducted by Bucher (1993). Male and female B6C3F1 mice were exposed to 1,3-butadiene for a single 2-hour period to concentrations of 0, 1000, 5000 or 10,000 ppm (2212, 11063 or 22126 mg/m³). The mice were then held for 2 years without treatment. There were no effects on survival at 2 years, no effects on bodyweight and no increased incidences of neoplastic or non-neoplastic lesions attributed to exposure to 1,3-butadiene. Although this study did not identify any carcinogenic effect associated with an acute exposure to 1,3-butadiene the studies with multiple exposures are more relevant for hazard assessment.

 

In conclusion, there is a marked species difference in the carcinogenicity of 1,3-butadiene in experimental animals. The difference in response is consistent with the in vivo genotoxicity of 1,3-butadiene in mice but not rats. In the mouse, 1,3-butadiene is a potent multi-organ carcinogen. Tumours develop after short durations of exposure, at low exposure concentrations and the carcinogenic response includes rare types of tumours. In the rat fewer tumour types, mostly benign develop at exposure concentrations of 100 to1000-times higher than in the mouse. The tumour response in rats suggests that non-genotoxic, possibly hormonal mechanisms influence the carcinogenicity in this species.

 

 

Human information

 

The European Union Risk Assesssment Report (EC, 2002) concluded that there is clear evidence from a study in styrene-butadiene rubber (SBR) workers (Delzell et al., 1995, 1996; Macaluso et al., 1996), that occupational exposure to 1,3-butadiene is associated with an excess of leukaemia. IARC (2008) concluded that there is sufficient evidence in humans for the carcinogenicity of 1,3-butadiene and noted that this conclusion was based primarily on the evidence for a significant exposure–response relationship between exposure to butadiene and mortality from leukaemia in an update of the SBR workers study (Sathiakumar et al., 2005; Graff et al., 2005: Delzell et al., 2006). More recently IARC have added a statement that “1,3-Butadiene causes cancer of thehaematolymphaticorgans” to their evaluation of the evidence from human studies (Baan et al., 2009). 

 

The SBR workers study provides good quality information on the association between exposure to 1,3-butadiene and haematolymphatic cancer (HLC) for a large group of over 16,000 workers with a long period of follow up, and the US EPA (2002) concluded that it provided the best published set of data to evaluate human cancer risk from 1,3-butadiene exposure, although the EU RAR (2002) stated that “overall these modelled data cannot be viewed as of sufficient reliability on which to base an estimate of the dose response relationship for the carcinogenic effect”. However, the most recent study update incorporates improved exposure estimates for 1,3-butadiene and estimates for potential confounders, styrene and dimethyldithiocarbamate (DMDTC), were calculated (Macaluso et al. 2004). The improved exposure estimates were validated by Sathiakumar et al. (2007). IARC (2008) discussed evidence for an association between butadiene and non-Hodgkin lymphoma which derives from the studies of workers in the monomer industry, and noted that they were unable to determine the strength of the evidence for particular histological subtypes of lymphatic and haematopoietic neoplasms because of changes in coding and diagnostic practices. The study by Divine and Hartman (2001) provides the most reliable information about the association between NHL and 1,3-butadiene exposure in monomer workers, but survival analyses showedno increase in risk with increasing cumulative1,3-butadieneexposurefor NHL and all HLC. Sathiakumar et al (2005) reported that they did not find any clear relation between employment in the SBR industry and other haematolymphatic cancers (besides leukaemia) and reported no excess of deaths from NHL and multiple myeloma.Graff et al.(2005) examined exposure-response trends for the same workers and reported a positive association between 1,3 -butadiene and leukaemia that was not explained by exposure to other agents examined, but they did not report similar associations for NHL and multiple myeloma.

 

Estimates of excess leukaemia risk have been derived using the SBR workers study by various groups (SCOEL, 2007; Sielken et al., 2007, 2008; TCEQ, 2008), but no estimates have been derived for HLC. Given the lack of association with other types of HLC, it seems unlikely that modelling the association between HLC and 1,3-butadiene in the SBR workers would provide better estimates of excess risk than modelling the association between leukaemia and 1,3-butadiene. Cheng et al (2007) used Cox regression procedures to examine the exposure–response relationship between several time-dependent 1,3-butadiene exposure indices and lymphoid neoplasms and myeloid neoplasms in addition to leukaemia. They concluded that evidence of an association between 1,3-butadiene and all lymphoid neoplasms or all myeloid neoplasms is less persuasive than that for all leukaemias. Graff et al. (2005) and Sielken et al. (2008) reported no association between acute myelogenous or monocytic leukaemia (AML) and 1,3-butadiene exposure, indicating that non-AML leukaemia may be a better endpoint than all leukaemia. However, there is limited information available to estimate the number of excess leukaemias of this type and quantitative risk assessment in the SBR workers cohort is based on models using all leukaemia as the endpoint.

 

SCOEL (2007) agreed that 1,3-butadiene should be treated as a possible human carcinogen, operating via a genotoxic mechanism. Excess risk entailed in exposure during a working life to various concentrations of 1,3-butadiene was calculated using a “step” approach (Zocchetti et al, 2004) for 23 sets of model parameters taken from Delzell et al. (2001). These dose response analyses for the SBR cohort incorporated the more refined exposure estimates of Macaluso et al. (2004), but not the additional 7 years of follow up of the latest study update. SCOEL estimated that occupational exposure to 1 ppm of BD for a working life (40 years between the ages of 25 and 65), will cause from 0.0 to 107.8 extra leukaemia deaths per 10⁴workers between the ages 25-85 years. However, 12 of the 23 SCOEL estimates are based on models which ignore exposure to BD at concentrations either below 100 ppm or above 100 ppm, and are not appropriate for risk assessment. With these 12 models excluded, the estimates for 1,3-butadiene exposure of 1 ppm range up to 15.3 per 10⁴excess deaths. However, the major weakness of the step model is that excess death estimates for low exposures are based on a single relative risk (RR) estimate which may have considerable variability. For example, the model giving the highest valid estimate of excess deaths at 1 ppm also gives the same estimate for all 1,3-butadiene exposures < 2.1 ppm, and these are based on a RR of 1.3 with 95% CI (0.4-4.3). In addition, the SCOEL approach doesn’t give a true range of estimates, especially for low exposures, as they are based on RR from different analyses which are highly correlated.

 

Sielken et al.(2008) used the SCOEL assumptions about the relevant exposure window and the same life table assumptions about mortality rates and survival probabilities as SCOEL (SCOEL, 2007) to calculate estimates of occupational risk. This report and an earlier report (Sielken et al.,2007) used Poisson and Cox regression models to model the association between 1,3-butadiene exposure and all leukaemias and leukaemia subtypes. Their models included terms for both cumulative 1,3-butadiene exposure and the cumulative number of exposures to 1,3-butadiene concentrations > 100 ppm (the number of High Intensity Tasks [HITs]). Their results show that cumulative BD HITS is an important predictor of risk with an effect that is independent of cumulative 1,3-butadiene exposure. The EU RAR (2002) had earlier noted that there was some indication that exposures accrued by exposure to 1,3-butadiene peaks may be important in the development of leukaemia, but there was insufficient data to clarify this. It can also be deduced from Graff et al (2005) that there were no leukaemia deaths among 34,152 person years of follow up from SBR workers exposed to 1,3-butadiene but not HITs. Sielken et al. (2008) noted that they considered Cox proportional hazards modelling to be more scientifically appropriate than Poisson regression modelling and Cox regression results are given more weight. They reported that all leukaemia and chronic myeloid leukaemia (CML) were associated with cumulative BD HITs, but not cumulative 1,3-butadiene exposure. Chronic lymphocytic leukaemia (CLL) was associated with cumulative 1,3-butadiene exposure, but not cumulative 1,3-butadiene HITs.Acute myelogenous or monocytic leukaemia (AML) was not associated with either cumulative 1,3-butadiene exposure or cumulative 1,3-butadiene HITs. For 1 ppm exposure and their preferred Cox regression model for leukaemia which adjusted for 1,3-butadiene HITs, Sielken et al (2008) estimated 0.33 extra leukaemia deaths per 10⁴workers. The corresponding estimate from a Poisson regression model that adjusted for 1,3-butadiene HITs was 0.53 extra leukaemia deaths per 10⁴workers. For CLL, the only endpoint significantly associated with cumulative 1,3-butadiene exposure before and after adjustment for 1,3-butadiene HITs, Sielken et al. (2008) derived an estimate of 0.16 extra CLL deaths per 10⁴workers.

 

Cheng et al. (2007) also modelled the leukaemia data from the SBR workers studies using a range of Cox regression models, and further analyses are included in the TCEQ (2008) report. TCEQ (2008) used life table methods to estimate general population lifetime risk estimates using the model parameters derived from these Cox regression models. These risk estimates are not relevant to an occupational exposure scenario, and the regression coefficients of the models are not directly comparable with those fitted by Sielken et al. (2008) which incorporated the SCOEL assumptions about the relevant exposure window and excluded exposures that occurred 40 or more years ago. Nevertheless, the magnitude of a regression coefficient relative to the coefficient of the Cox regression model fitted by Sielken et al. (2008) for which occupational risk estimates are available (continuous cumulative 1,3-butadiene exposure with age and number of 1,3-butadiene HITs > 100 ppm as covariates), can be used to give an approximate estimate of the number of excess leukaemias. Cheng et al. (2007) noted that the slope of the exposure–response relationship was irregular especially in the upper part of the 1,3-butadiene exposure range. In order to reduce the impact of data in the upper part of the cumulative 1,3-butadiene exposure range, Cheng et al. (2007) fitted Cox regression models using either continuous cumulative BD exposure restricted to the lower 95% of exposure range, or the mean scored deciles of cumulative 1,3-butadiene exposure. However, Cheng et al (2007) noted that they preferred the estimate of the exposure–response trend that is based on the continuous, untransformed form of the 1,3-butadiene variables and the full range of exposure data, but noted the high potential for distortion of the exposure–response relationship as a result of exposure misclassification. Table 1 shows Cox regression coefficients for models in which the hazard function was either a log linear function of continuous cumulative 1,3-butadiene exposure or a log linear function of the mean scored deciles of cumulative BD exposure. In addition, the table also shows Cox regression coefficients for the continuous model using the exposure metric that excluded exposure that occurred more than 40 years ago or excluded the 5% of workers with the highest cumulative 1,3-butadiene exposures. Regression coefficients are also shown for models which were fitted with and without BD HITs as covariate. There is little difference between the continuous and categorical regression models that adjust for 1,3-butadiene HITs and do not exclude exposure that occurred more than 40 years ago. All of the regression coefficients were close to that of the preferred model of Sielken et al. (2008) except for that of the mean scored deciles model that did not adjust for 1,3-butadiene HITs. The coefficient of the Poisson regression model derived by Sielken et al. (2008) that did not adjust for 1,3-butadiene HITs (1.76 x 10^-3) was also much higher than the regression model that did adjust for BD HITs (3.42 x 10^-4), and is also based on mean scored deciles of cumulative BD exposure. 

 

Table 1 Values of Maximum Likelihood Estimate ofβ, standard error (SE) and size of coefficient relative to that of the Sielken et al.(2008) Cox regression model with adjustment for1,3-butadieneHITs

Model	Covariates
	Age		Age & Number of 1,3-butadiene HITs > 100 ppm
	Source	β (MLE)±SE	Source	β (MLE)±SE
Cox log-linear ppm-years continuous, excluding exposure that occurred > 40 years ago	Sielkenet al.(2008)	2.93E-04 ± 1.05E-04 1.4 fold higher^2	Sielkenet al.(2008)	2.15E-04 ± 1.31E-04 Baseline^1
Cox log-linear ppm-years continuous	Chenget al.(2007)	2.9E-04 ± 1.0E-04 1.4 fold higher^2	Chenget al.(2007)	2.5E-04 ± 1.2E-04 1.2 fold higher^2
Cox log-linear ppm-years mean-scored deciles	Chenget al.(2007)	7.5E-04 ± 2.2E-04 3.5 fold higher^2	TCEQ (2008)	2.8E-04 ± 2.4E-04 1.3 fold higher^2

¹Baseline model is the preferred Cox regression model of Sielken et al. (2008) with adjustment for HITs

²Relative to the coefficient of the baseline model

 

For occupational exposure, a DMEL of 1 ppm (2.21mg/m³) is proposed. The estimate of excess leukaemia deaths(all cell types combined)derived by Sielken et al (2008) is 0.33 x 10^-4, and is less than 0.4 x 10^-4, which has been proposed as a future limit for acceptable occupational risk (AGS, 2008). Regression coefficients from other Cox regression models reported by Cheng et al (2007) and TCEQ (2008), and estimates from Poisson regression models, indicate that other risk estimates are generally close to 0.4 x 10^-4, even if based on regression models that do not adjust for 1,3-butadiene HITs. All of the estimates are considerably lower than the current limit for acceptable occupational risk of 4 x 10^-4that has recently been proposed (AGS,2008). In addition, the estimate derived by Sielken et al (2008) is very precise with an upper bound of 0.66 x 10^-4based on a one-sided 95% upper confidence limit (UCL) for the regression parameter (the slope parameter was not statistically significantly different to zero). In addition, the risk estimate derived from using CLL as an endpoint, suggests that 1 ppm may be conservative.

 

With regard to general population exposure, Sielken et al (2007) derived estimates of excess leukaemia deaths using a Poisson regression model which adjusted for1,3-butadieneHITs. TCEQ (2008) derived estimates using this model and also for various Cox regression models fitted by Cheng et al (2007), with and without adjustment for1,3-butadieneHITs.TCEQ (2008) calculated 10^-5risk air concentrations by extrapolation linearly from the 10^-3risk air concentration derived using life-table methods. Most estimatesalso incorporated an age-adjustment factor to reflect the potential for early-life exposure to make a greater contribution to cancers appearing later in life. Both Sielken et al. (2008) and Cheng et al (2007) noted that Cox regression techniques have several advantages over Poisson regression models, and the estimates derived from Cox regression models are given precedence. The model selected by TCEQ (2008) for risk assessmentwas the Cox regression model for continuous cumulative 1,3-butadiene exposure restricted to the lower 95% of exposure range. Table 2 shows estimates of the air concentrations corresponding to 1 in 10⁵extra leukemia risk for this Cox regression model, and those for continuous cumulative BD exposure over the whole exposure range of subjects and mean scored deciles of cumulative 1,3-butadiene exposure. Estimates are shown for models which were fitted with and without 1,3-butadiene HITs as a covariate.

 

Table 2Ambient airconcentrations corresponding to 1 in 10⁵extra leukemia risk for Cox regression models with and without adjustment for1,3-butadiene HITs (taken from TCEQ [2008])

Model	Age	Age & Number of 1,3-butadiene HITs > 100 ppm
	Air Concentration 1 in 10^5excess cancer risk using model	Air Concentration 1 in 10^5excess cancer risk using model
Cox log-linear ppm-years continuous	69.79 ppb	80.95 ppb
Cox log-linear ppm-years mean-scored deciles	26.98 ppb	72.28 ppb
Cox log-linear (restricted to lower 95% of exposure range) ppm-years continuous	12.81 ppb	15.10 ppb

 

Although TCEQ (2008) based their analysis on theCox regression model for continuous cumulative 1,3-butadiene exposure restricted to the lower 95% of exposure range, there does not seem to be any good reason to give this model any more weight than the other two models, and the mean scored deciles model may be a better approach to assess the impact of data in the upper part of the cumulative 1,3-butadiene exposure range. The geometric mean of the three estimates derived from models that did not adjust for 1,3-butadiene HITs is 28.9 ppb (30.1 ppb without use of the additional age-adjustment factor to reflect higher potential risk at early ages). The geometric mean of the three estimates derived from models that did adjust for 1,3 -butadiene HITs is 44.5 ppb (no estimates available without use of the age-adjustment factor).Estimates based on the Poisson linear regression model with mean scored deciles (Sielken et al., 2007) behaved similarly to the Cox log-linear model with mean scored deciles and do not change the picture. However, the range of air concentrations giving a 1 in 10⁵risk was much wider than for the equivalent Cox regression model; 14.33 ppb (age only) to 127.4 ppb (age & number of 1,3-butadiene HITs > 100 ppm).  

The inclusion of age and number of HITs > 100 ppm 1,3-butadiene as covariates in the Cox regression modeling may result in cancer potency estimates that are more relevant to 1,3-butadiene exposures experienced by the general population. The general population is not expected to be exposed to 1,3-butadiene concentrations greater than 100 ppm, so adjusting for 1,3-butadiene HITs > 100 ppm as a covariate produces cancer potency estimates more relevant to BD exposures experienced by the general population. However, 1,3-butadiene ppm years and number of HITs may be correlated and it may not be appropriate to include both of them in the same model. Sielken (2008) concluded that the effects of the cumulative number of 1,3-butadiene HITs and cumulative 1,3-butadiene ppm-years were independent, and were successfully disentangled in the Poisson regression model (Sielken et al., 2007), and further concluded that this was expected to be even more true for the Cox regression model used to estimate occupational risk because the model used individual values for cumulative 1,3-butadiene ppm-years rather than group values. Cheng et al. (2007) noted that these 1,3-butadiene exposure variables were weakly correlated for continuous values, but that grouped (deciles) values were more highly correlated (Pearson correlation coefficient of 0.80). From Table 21 it can be seen that adjusting for 1,3-butadiene HITs only made an appreciable difference for the Cox regression model with mean-scored deciles of cumulative 1,3-butadiene exposure. As this might have resulted from correlation between the two 1,3-butadiene exposure variables, the estimates derived from models that did not adjust for 1,3-butadiene HITs were selected for general population risk assessment. A DMEL of 30 ppb (0.0664 mg/m³) is proposed based on the geometric mean of the risk estimates derived from Cox regression models that did not adjust for peak exposures.

  

Additional references

 

Baan R, Grosse Y, Straif K, Secretan B, El Ghissassi F, Bouvard V, Benbrahim-Tallaa L, Guha N, Freeman C, Galichet L and Cogliano V; WHO International Agency for Research on Cancer Monograph Working Group. (2009). A review of human carcinogens--Part F: chemical agents and related occupations. Lancet Oncol. 10 (12),1143-4.

Committee on Hazardous Substances (AGS). (2008). Guide for the quantification of cancer risk figures after exposure to carcinogenic hazardous substances for establishing limit values at the workplace. 1. Edition. Dortmund: Bundesanstalt für Arbeitsschutz und Arbeitsmedizin. Availablehttp://www.baua.de/cae/servlet/contentblob/717582/publicationFile/48510/Gd34.pdfandhttp://www.baua.de/cae/servlet/contentblob/665100/publicationFile/48349/Announcement-910.pdf

Delzell E, Macaluso M, Sathiakumar NandMatthews R. (2001). Leukemia and exposure to 1,3-butadiene, styrene and dimethyldithiocarbamate among workers in the synthetic rubber industry.Chem Biol Interact, 135-136, 515-34.

Delzell, E, N Sathiakumar, and M Hovinga. (1996). A follow-up study of synthetic rubber workers. Toxicology 113, 182-189.

Delzell, E, N Sathiakumar, and M Macaluso. (1995). A follow-up study of synthetic rubber workers. Final report prepared under contract to International Institute of Synthetic Rubber Producers.

ECETOC (1997). 1,3-Butadiene OEL Criteria document. Special Report No. 12

EU RAR (2002). European Union Risk Assessment Report for 1,3-butadiene. Vol. 20. European Chemicals Bureau (http: //ecb. jrc. ec. europa. eu/DOCUMENTS/Existing-Chemicals/RISK_ASSESSMENT/REPORT/butadienereport019. pdf)

International Agency for Research on Cancer (IARC). (2008). IARC Monographs on the evaluation of carcinogenic risks to humans. Volume 97. 1,3-Butadiene, ethylene oxide, and vinyl halides (vinyl fluoride, vinyl chloride and vinyl bromide). Lyon: International Agency for Research on Cancer. pp. 45-185.

Macaluso, M, Larson, R, Delzell, E, Sathiakumar, N, Hovinga, M, Julian, J, Muir, D and Cole, P. (1996) Leukemia and cumulative exposure to butadiene, styrene and benzene among workers in the synthetic rubber industry. Toxicology, 113, 190-202.

Macaluso, M, Larson, R, Lynch, J, Lipton, S and Delzell, E. (2004). Historical estimation of exposure to 1,3-butadiene, styrene, and dimethyldithiocarbamate among synthetic rubber workers. J. Occup. Environ. Hyg. 1, 371-390.

Sathiakumar N, Delzell., Cheng H, Lynch J, Sparks W and Macaluso M. (2007). Validation of 1,3-butadiene exposure estimates for workers at a synthetic rubber plant, Chem. -Biol. Interact. 166, 29–43.

SCOEL. (2007). Recommendation from the Scientific Committee on Occupational Exposure Limits: risk assessment for 1,3-butadiene. SCOEL/SUM/75 final (updated Feb 2007).

Texas Commission on Environmental Quality (TCEQ) (2008). Development Support Document. 1,3-Butadiene. Chief Engineer’s Office. Available: http: //tceq. com/assets/public/implementation/tox/dsd/final/butadiene, _1-3-_106-99-0_final. pdf

United States Environmental Protection Agency (USEPA). (2002). Health Assessment of 1,3-Butadiene. EPA/600/P-98/001F. National Center for Environmental Assessment, Office of Research and Development, Washington D. C.

Zocchetti C, Pesatori AC, Bertazzi PA.(2004) [A simple method for risk assessment and its application to 1,3-butadiene].Med Lav. 95(5), 392-409.