Priority Substances List Assessment Report- 1,3-Butadiene

3.0 Assessment of "Toxic" Under CEPA 1999 (Continued)

3.3 CEPA 1999 64(c): Human health

3.3.3.1 Carcinogenicity

3.3.3.1.2 Estimated potency based on data from studies in experimental animals

As described in Section 3.3.2, butadiene induced an increase in the incidence of tumours at multiple sites in both B6C3F₁ mice (liver, lung, Harderian gland, mammary gland, ovaries, forestomach, Zymbal gland and kidney, along with malignant lymphomas, histiocytic sarcomas and cardiac hemangiosarcomas) and Sprague-Dawley rats (mammary gland, thyroid gland, uterus, Zymbal gland, pancreas and testes). As discussed above, consistent with the species differences in metabolism, mice were much more sensitive to butadiene-induced cancer than were rats for the strains investigated. Based on data available (i.e., evidence from genotoxicity studies that butadiene and its metabolites are active in both species), this difference in sensitivity is quantitative rather than qualitative and is related to the greater amounts of putatively active metabolites formed in mice compared with rats. In addition, the different profiles of tumours observed in the two species may be related to differential roles of the epoxide metabolites in the induction of the various tumours; i.e., the diepoxide may be more critical to tumour induction in mice than is EB (since it was reported recently that formation of DEB increased with level of exposure to butadiene in mice but not in rats; Thornton-Manning et al., 1998), while the monoepoxide or monoepoxide diol may be more important in rats.

The relevance for extrapolation to humans of exposure-response for some of the types of tumours observed in rodents has been questioned. For example, Irons et al. (1989) hypothesized that the thymic lymphoma/leukemia induced in B6C3F₁ mice may be related to the presence of an endogenous ecotropic retrovirus, as a much lower incidence was observed in Swiss mice that do not possess this retrovirus (although the incidence was significantly elevated compared with controls). Therefore, although the hematopoietic system is a target for the induction of cancer by butadiene in humans, the observed exposure-response relationship for this endpoint is not considered appropriate for quantitative extrapolation to humans - on the basis that this retrovirus is not present in humans and its presence in B6C3F₁ mice renders this strain quite susceptible to induction of lymphoma - although the relevant information is included for comparative purposes.

It has also been suggested that the tumours observed in the study in rats (i.e., mammary gland, thyroid gland, pancreas, uterus and testes) and some of the tumours induced in mice (i.e., ovaries and mammary gland) may be mediated through effects on the endocrine system. Indeed, tumours at these sites are often associated with disruption of hormonally mediated functions. In addition, non-neoplastic or pre-neoplastic effects, including atrophy, degeneration and hyperplasia, have also been observed in mice exposed subchronically to butadiene. However, the mechanism by which butadiene induces tumours at these sites has not yet been adequately investigated; i.e., it has not been established whether these tumours are induced via a mechanism for which there may be a threshold of exposure (e.g., through induction of hormonally mediated effects), although the possibility is recognized. In addition, the results of in vivo genotoxicity assays indicate that butadiene or its metabolites induce genetic effects in the reproductive organs of multiple strains of mice.

Based on these considerations, estimates of carcinogenic potency were calculated on the basis of the malignant lymphomas, histiocytic sarcomas, cardiac hemangiosarcomas, alveolar/bronchiolar adenomas or carcinomas, hepatocellular adenomas or carcinomas, squamous cell papillomas or carcinomas of the forestomach, adenomas or carcinomas of the Harderian gland, granulosa cell tumours of the ovaries and adenoacanthomas, carcinomas or malignant mixed tumours of the mammary gland observed in B6C3F₁ mice in the chronic bioassay conducted by the NTP (1993) and the mammary gland tumours, pancreatic exocrine adenomas, Leydig cell tumours, Zymbal gland carcinomas, thyroid follicular cell adenomas or carcinomas and uterine sarcomas in Sprague-Dawley rats reported by Hazleton Laboratories Europe Ltd. (1981a). (The tumour incidence data for each of the sites considered are presented in Table 3.) It is noted that the characterization of exposure-response is much better in the study in mice (which involved five closely spaced exposure levels) than in the bioassay in rats (in which only two more widely spaced exposure levels were used, the higher of which was likely above the level of metabolic saturation). (N.B.: Although there were also increased incidences of tumours at several sites in B6C3F₁ mice in the "stop-exposure" study conducted by the NTP [1993], only TC₀₅s determined on the basis of the 2-year study were included, as the latter study provides better information for characterization of exposure-response in mice following long-term exposure [i.e., more exposure levels for up to 2 years].)

In the NTP study, mice were exposed to 0, 6.25, 20, 62.5, 200 or 625 ppm (0, 13.8, 44.2, 138, 442 or 1383 mg/m³) butadiene for 6 hours per day, 5 days per week, for 103 weeks. Survival of mice decreased with increasing exposure concentration; therefore, to minimize the effect of the high mortality rate, the poly-3 adjusted data presented in the NTP (1993) report were used in these calculations. For some tumour types, the adjusted data still demonstrated downward curvature at the highest concentration. In these cases, the high-exposure group was excluded in the determination of the TC₀₅. The TC₀₅s were calculated for these endpoints by first fitting a multistage model to the data. The multistage model is given by:

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

The models were fitted using GLOBAL82 (Howe and Crump, 1982), and a chi-square lack of fit test was performed for each model fit. The degrees of freedom for this test are equal to k minus the number of q_i's whose estimates are non-zero. A p-value less than 0.05 indicates a significant lack of fit. Results from the model fitting are displayed in Table 12. Plots of the data and the fitted models are shown in Figure 3.

TC₀₅s were determined as the doses D (in mg/m³) that satisfy

and then adjusted by multiplying by:

where, in the first term, which amortizes the dose to be constant over the lifetime of a mouse, w is the duration of the experiment (103 weeks). The second factor was suggested by Peto et al. (1984) and corrects for an experiment length that is unequal to the standard lifetime. Since tumours develop much more rapidly later in life, a greater than linear increase in the tumour rate is expected when animals are observed for tumours longer than their standard lifetime (or the reverse when animals are observed for a period shorter than their standard lifetime). (N.B.: Application of this factor does not impact greatly on the final values, since it is very close to one.) The selected TC₀₅ values for this study and their 95% lower confidence limits (LCLs) are presented in Table 12 and range from 2.3 mg/m³ (95% LCL = 1.7 mg/m³) or 1.1 ppm (95% LCL = 0.79 ppm) for Harderian gland tumours in males to 99 mg/m³ (95% LCL = 23 mg/m³) or 45 ppm (95% LCL = 10 ppm) for malignant lymphomas in males.

Table 12 Carcinogenic potency estimates (TC₀₅s) 1 of butadiene based on results of bioassays in experimental animals

Enlarge image

Figure 3 Exposure-response analysis for butadiene-induced tumours in mice

*TC₀₅ unadjusted for lifetime dosing

Estimates of carcinogenic potency were also calculated based on the results of the bioassay in Sprague-Dawley rats (Hazleton Laboratories Europe Ltd., 1981a). In this study, rats were exposed to 0, 1000 or 8000 ppm (0, 2212 or 17 696 mg/m³) for 6 hours per day, 5 days per week, for 1₀₅ (males) or 111 (females) weeks. A high mortality rate was observed at the higher concentration; therefore, this exposure group was excluded from the analysis, except for the potency estimates for pancreatic exocrine adenomas in males (for this endpoint, exclusion of the high-exposure group would have resulted in the exposure-response relationship curving downwards). As for mice, a multistage model was fit to the data for rats using GLOBAL82 and adjusted to account for study duration (w) by multiplying by:

where the duration of the experiment was 105 weeks for males and 111 weeks for females.

The exposure-response curves and estimated adjusted TC₀₅ values based on this study in rats are presented in Figure 4 and Table 12, respectively. The concentrations of butadiene estimated to be associated with a 5% increased incidence of tumours ranged from 6.7 mg/m³ (95% LCL = 4.7 mg/m³) or 3.0 ppm (95% LCL = 2.1 ppm) to 4872 mg/m³ (95% LCL = 766 mg/m³) or 2203 ppm (95% LCL = 346 ppm) for tumours of the mammary gland and Zymbal gland in female rats, respectively. Although the available data for analysis of exposure-response were more limited for rats than for mice, it is interesting to note the similarity in estimates of potency for mammary gland tumours (i.e., 6.7 mg/m³ in both species).

Based on modelling (using THC; Howe, 1995a) of the incidence of micronucleated polychromatic erythrocytes in B6C3F₁ mice exposed to butadiene for up to 15 months in the NTP bioassay, BMC₀₅s for somatic cell mutations were very similar to the lower end of the range of the TC₀₅s for tumour induction.

3.3.3.2 Non-neoplastic effects

There have been recent attempts to quantitatively estimate risk of heritable genetic damage in humans based on a parallelogram approach and data on male-mediated heritable translocations and bone marrow micronuclei in mice and chromosomal aberrations in lymphocytes of exposed workers (Pacchierotti et al., 1998b). In view, however, of the reported ovarian atrophy due to reduction of primordial follicles (to a degree that would preclude reproduction) following chronic exposure of mice to concentrations of butadiene considerably lower than those associated with adverse effects on the testes, investigation of the response of female germ cells in mice to butadiene is desirable, since this may well be the most sensitive endpoint for development of quantitative estimates of heritable damage. (Determination of putatively toxic metabolites in the ovaries of butadiene-exposed female mice would also be informative.) For this reason, quantitation of exposure-response for heritable genetic damage is not presented here. However, in view of the apparent greater sensitivity of the reproductive organs in female mice, a benchmark concentration was derived for non-neoplastic effects in the ovary, which is considerably more protective than that for male-mediated heritable damage developed by Pacchierotti et al. (1998b). (Although the relative role of butadiene in the induction of the observed atrophy in mice in the NTP study is unclear, as discussed in Section 3.3.2.2, information currently available is not considered a sufficient basis upon which to dismiss this endpoint as being inappropriate for quantification of exposure-response. However, this uncertainty should be kept in mind in the interpretation or application of the BMC₀₅s derived below.)

Figure 4 Exposure-response analysis for butadiene-induced tumours in rats

*TC₀₅ unadjusted for lifetime dosing

Hematotoxicity is considered to be a critical effect associated with exposure to butadiene. Although the hematopoietic system appears to be a target for butadiene-induced cancer in humans, available data on the potential non-neoplastic effects on this system are inadequate for quantitation of exposure-response. However, since statistically significant changes were observed in mice only at concentrations greater than those that induced other toxic effects, and since benchmark concentrations derived for effects on the blood are greater than those for these other effects, quantitation of the exposure-response for hematological effects has not been presented here.

Ovarian atrophy was observed in both long-term NTP (1984, 1993) bioassays in mice and a subchronic study (Bevan et al., 1996). Although limited, available data indicate that rats are less sensitive to induction of this effect, which may, again, be a consequence of interspecies variations in metabolism. Therefore, although additional research into the etiology of the observed ovarian atrophy in mice would be desirable, the data from the later NTP study are considered most appropriate for characterization of exposure-response (i.e., development of a BMC₀₅). In this investigation, the incidence of atrophy of the ovaries was significantly increased in an exposure-related manner at all concentrations tested (i.e., ≥ 6.25 ppm [≥13.8 mg/m³]). The severity of this effect also increased with exposure (see Table 13).

The exposure-response relationship for ovarian atrophy from this study was quantified by fitting the following model to the dose-response data (Howe, 1995b):

where d is dose, k is the number of dose groups in the study minus one,
P(d) is the probability of the animal developing the effect at dose d and
q_i> 0, i =1,..., k and d0 are parameters to be estimated. The models were fit using THRESH (Howe, 1995b), and the BMC₀₅s were calculated as the dose D that satisfies:

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

The BMC₀₅ was then amortized to be constant over the standard life of a mouse by multiplying by:

Resulting BMC₀₅s and lack of fit information for all models fit are displayed in Table 14.

The model fitted to all six exposure groups exhibited a significant lack of fit, likely due to the fact that the curve rises sharply and then plateaus at the three highest exposure groups. Plots of the data and fitted model are displayed in Figure 5. Since a good fit in the range of the BMC₀₅ (in the vicinity of 6.25 ppm [13.8 mg/m³]) is desired, the model was refitted omitting the two highest exposure groups. This model again indicates a marginal lack of fit. The graph of this model (Figure 6) indicates that this model provides a reasonable visual fit to the data, but the resulting BMC₀₅ is uncertain due to lack of fit of the model.

Table 13 Incidence and severity of ovarian atrophy observed in 2-year bioassay in mice (NTP, 1993)

Enlarge image

Figure 5 Exposure-response analysis for ovarian atrophy in mice

*BMC₀₅ and BMCL₀₅ unadjusted for lifetime dosing

The BMC₀₅ for the model excluding the two highest dose groups was calculated to be 0.57 mg/m³, with a 95% LCL of 0.44 mg/m³.

Figure 6 Exposure-response analysis for ovarian atrophy in mice, excluding two highest dose groups

*BMC₀₅ and BMCL₀₅ unadjusted for lifetime dosing

Table 14 Benchmark concentrations for ovarian atrophy

Ovarian atrophy	BMC05 (ppm)	95% LCL on BMC05 (ppm)	BMC05 (mg/m3)	95% LCL on BMC05 (mg/m3)	Chi- square	df	p-value
All severities	2.5	1.9	5.6	4.1	61	4	0.00
All severities, excluding top two dose groups	0.25	0.20	0.57	0.44	7.0	2	0.03
Moderate/marked severity	4.3	3.4	9.6	7.6	37.1	4	0.00
Moderate/marked severity, excluding top dose group	1.4	1.1	3.1	2.5	2.2	3	0.55

Figure 7 Exposure-response analysis for moderate/marked ovarian atrophy

*BMC₀₅ and BMCL₀₅ unadjusted for lifetime dosing

Figure 8 Exposure-response analysis for moderate/marked ovarian atrophy, excluding high-dose group

*BMC₀₅ and BMCL₀₅ unadjusted for lifetime dosing

If only those animals that had moderate or marked ovarian atrophy from all exposure groups were included, the resulting BMC₀₅ would be 9.6 mg/m³ (95% LCL = 7.6 mg/m³), although there is again a significant lack of fit (Figure 7). If the highest exposure group is excluded, the BMC₀₅ for moderate or marked ovarian atrophy becomes 3.1 mg/m³, with a 95% LCL of 2.5 mg/m³ (Figure 8).

3.3.4 Human health risk characterization

Butadiene is released to air in Canada from both industrial point sources and more dispersive, non-point sources, the latter due to its production primarily during incomplete combustion. Intake for the general population in Canada is primarily from air, with intake from other media likely being negligible in comparison. The focus of the human health risk characterization is, therefore, the general population exposed in outdoor and indoor air in the general environment and those exposed through air in the vicinity of industrial point sources.

For compounds such as butadiene, where data are sufficient to support a plausible mode of action for induction of tumours by direct interaction with genetic material, estimates of exposure are compared with quantitative estimates of cancer potency (Exposure Potency Index or EPI) to characterize risk and provide guidance in establishing priorities for further action (i.e., analysis of options to reduce exposure) under CEPA (Health Canada, 1994).

Tumorigenic concentrations were calculated on the basis of data from both epidemiological studies and investigations in experimental animals. For the critical epidemiological investigation (Delzell et al., 1995), a TC₀₁ (i.e., the concentration associated with a 1% increase in mortality due to leukemia) was considered the appropriate measure of carcinogenic potency, since the majority of the observable data fell within this range. Although four different mathematical models were considered, the TC₀₁ generated by the model with the best fit was 1.7 mg/m³.

Quantitative estimates of carcinogenic potency derived on the basis of data in experimental animals were calculated as TC₀₅s (i.e., the concentration associated with a 5% increase in tumour incidence). Based on the 2-year bioassay in mice (NTP, 1993), TC₀₅s ranged from 2.3 mg/m³ (95% LCL = 1.7 mg/m³) to 99 mg/m³ (95% LCL = 23 mg/m³). The TC₀₅s derived on the basis of the more limited study in rats (Hazleton Laboratories Europe Ltd., 1981a) ranged from 6.7 mg/m³ (95% LCL = 4.7 mg/m³) to 4872 mg/m³ (95% LCL = 766 mg/m³).

The values derived on the basis of studies in humans are preferred as the basis for comparison with estimates of exposure to characterize risk. While there are a number of uncertainties in the use of the epidemiological data for both hazard evaluation and exposure- response analyses (Section 3.3.5), these are likely far less than uncertainties associated with interspecies extrapolation. Moreover, estimated potency for humans is similar to that developed on the basis of the cancer bioassays in experimental animals. (Indeed, although in an area of the exposure-response curve where data were more sparse, it is noteworthy that TC₀₅s calculated on the basis of epidemiological data [as opposed to the TC₀₁s presented above] are within the range of values derived from the studies in rodents.)

Based on the estimates of exposure presented above (Section 3.3.1), 95% of the population is exposed to concentrations of butadiene in outdoor air of 1.0 µg/m³ or less. For the proportion of the general population that is regularly exposed to higher concentrations of butadiene in urban areas (i.e., the "reasonable worst-case scenario"), the 95th percentile of the distribution of concentrations is 1.3 µg/m³. In the only area of Canada identified as having an industrial point source, the 95th percentile of the distribution of concentrations is 6.4 µg/m³.

Comparison of Estimates of Carcinogenic Potency with Exposure Levels

Exposure	Potency (TC01 or TC05)	Margin between effect level and exposure*	Priority for further action*
		Exposure Potency Index (EPI)*
1.0 µg/m3 (95th percentile for all sites in Canada)	1.7 mg/m3 (TC01 for leukemia in humans)	1700	Moderate
		5.9 × 10-4
	2.3 mg/m3 (TC05 for most sensitive tumour site in mice [Harderian gland])	2300	High
		4.3 × 10-4
	1.7 mg/m3 (95% LCL of TC05 for most sensitive tumour site in mice)	1700	High
		5.9 × 10-4
	6.7 mg/m3 (TC05 for most sensitive tumour site in rats [mammary gland])	6700	Moderate
		1.5 × 10-4
	4.7 mg/m3 (95% LCL of TC05 for most sensitive tumour site in rats)	4700	High
		2.1 × 10-4
1.3 µg/m3 (95th percentile for reasonable worst-case scenario)	1.7 mg/m3 (TC01 for leukemia in humans)	1300	Moderate
		7.6 × 10-4
	2.3 mg/m3 (TC05 for most sensitive tumour site in mice [Harderian gland])	1800	Hight
		5.7 × 10-4
	1.7 mg/m3 (95% LCL of TC05 for most sensitive tumour site in mice)	1300	High
		7.6 × 10-4
	6.7 mg/m3 (TC05 for most sensitive tumour site in rats [mammary gland])	5200	Moderate
		1.9 × 10-4
	4.7 mg/m3 (95% LCL of TC05 for most sensitive tumour site in rats)	3600	High
		2.8 × 10-4
6.4 µg/m3 (95th percentile for area affected by industrial point source)	1.7 mg/m3 (TC01 for leukemia in humans)	270	High
		3.8 × 10-3
	2.3 mg/m3 (TC05 for most sensitive tumour site in mice [Harderian gland])	360	High
		2.8 × 10-3
	1.7 mg/m3 (95% LCL of TC05 for most sensitive tumour site in mice)	270	High
		3.8 × 10-3
	6.7 mg/m3 (TC05 for most sensitive tumour site in rats [mammary gland])	1000	High
		9.6 × 10-4
	4.7 mg/m3 (95% LCL of TC05 for most sensitive tumour site in rats)	730	High
		1.4 × 10-3

* For EPIs calculated on the basis of a TC₀₁ derived from epidemiological data, the priority for investigation of options to reduce exposure is considered to be high, moderate or low if the EPI values are determined to be 1 × 10^-3 or greater, between 1 × 10^-5 and 1 × 10^-3 or less than 1 × 10^-5, respectively. If EPIs are calculated on the basis of a TC₀₅ derived from data in laboratory animals, the priority for investigation of options to reduce exposure is considered to be high, moderate or low if the EPI values are determined to be 2.0 × 10^-4 or greater, between 2.0 × 10^-6 and 2.0 × 10^-4 or less than 2.0 × 10^-6, respectively.

The margins between carcinogenic potency and estimated exposure for the general population (including ambient and reasonable worst case) and those in the vicinity of a point source are presented in the table above. Based on these margins, the priority for investigation of options to reduce exposure of the general population exposed in the ambient environment is considered to be moderate to high, while that for those in the vicinity of industrial point sources is considered to be high.

In view of the relative potency of butadiene to induce some non-cancer effects, these endpoints are also important in risk characterization. As presented above, a benchmark concentration (BMC₀₅) of 0.57 mg/m³ (95% LCL = 0.44 mg/m³) was derived on the basis of data for the incidence of ovarian atrophy of all severities (i.e., female reproductive toxicity) in mice exposed to butadiene for up to 2 years (NTP, 1993). And while there is uncertainty about the relevance of the ovarian atrophy observed in mice for humans (Section 3.3.5), the BMC₀₅ is slightly less than the lower end of the range of estimates of cancer potency based on the incidence of tumours in the same study in mice, as well as the TC₀₅ for cancer based on the epidemiological data. The mode of induction of ovarian atrophy is unknown. However, if it is (reasonably) assumed that the mode of action is related to that by which tumours are induced (i.e., direct interaction with genetic material), the priority for further action, based on the margin between estimated potency and exposure, is considered to be high. It should be noted, though, that even if the mode of induction of ovarian atrophy does not involve direct interaction with genetic material, the margin between exposure and effect level (i.e., for which a tolerable concentration is normally developed) is still inadequate - i.e., exposure levels in Canada are 90-570 times lower than the benchmark concentration, as presented below. Therefore, the priority for further action (i.e., priority for investigation of options to reduce exposure), based on this effect, is considered to be high.

Based on comparison of estimated exposure with the potency to induce leukemia in humans and cancer and non-cancer effects in experimental animals, and taking into consideration the degree of confidence in the database upon which the quantitative measures of toxicity were based, the overall priority for investigation of options to reduce exposure to butadiene in the general environment in Canada, based solely on potential adverse health effects, is considered to be moderate to high.

Comparison of Estimates of Potency for non Cancer Effects with Exposure Levels

Exposure	Potency (BMC05)	Margin between effect level and exposure*	Priority for further action*
		Exposure Potency Index (EPI)*
1.0 µg/m3 (95th percentile for all sites in Canada)	0.57 mg/m3 (BMC05 for ovarian atrophy in mice)	570	High
		1.8 × 10-3
	0.44 mg/m3 (95% LCL of BMC05 for ovarian atrophy in mice)	440	High
		2.3 × 10-3
1.3 µg/m3 (95th percentile for reasonable worst-case scenario)	0.57 mg/m3 (BMC05 for ovarian atrophy in mice)	440	High
		2.3 × 10-3
	0.44 mg/m3 (95% LCL of BMC05 for ovarian atrophy in mice)	340	High
		3.0 × 10-3
6.4 µg/m3 (95th percentile for area affected by industrial point source)	0.57 mg/m3 (BMC05 for ovarian atrophy in mice)	90	High
		1.1 × 10-2
	0.44 mg/m3 (95% LCL of BMC05 for ovarian atrophy in mice)	70	High
		1.5 × 10-2

* If mode of action involves interaction with genetic material.