- Email: [email protected]
Risk assessment of butadiene in ambient air; the approach used in the UK
ELSEVIER
Toxicology 113 (1996) 221-225
Risk assessment of butadiene in ambient air; the approach used in the UK1 R.J. Fielder Department oj Health (HEF Division), Skipton House, Elephant and Castle, London SE1 6L W, UK
The UK Committee on Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (COC) and the related Committee on Mutagenicity provided advice on 1,3-butadiene in 1992. This followed detailed consideration of the available mutagenicity, animal carcinogenicity and epidemiology data plus information on toxicokinetics. They concluded that 1,3-butadiene was an in vivo mutagen, a potent genotoxic animal carcinogen and should be regarded as a probable human carcinogen. The Department of Health is not aware of more recent data warranting reconsideration of these conclusions. General advice on setting air quality standards for carcinogenic air pollutants was given by the COC. Although the prudent assumption of the absence of any safe level for genotoxic carcinogens was preferred, a pragmatic approach based essentially on assessment of the exposure at which no increased risk would be detected, plus a safety factor, was considered reasonable for compounds like butadiene where exposure cannot be totally avoided. This approach, plus recognition that it is unadvisable to allow ambient levels of genotoxic carcinogens to rise, is used in the UK. The procedure by which the Department of Environment’s Expert Panel on Air Quality Standards recommended a value of 1 ppb for butadiene based on these principles is described. Keywords:
1,3-Butadiene; Genotoxic carcinogen; Air quality standard
1. Introduction In the UK, the Department of Health (DH) has established the Committee on Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (COC) and also the Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment (COM). Their charge is to advise Government ‘The opinions presented in this paper are those of the author and in no way commit the Department of Health or other Government Departments.
Departments and Agencies on the carcinogenic or mutagenic risk of chemicals, and on the general principles involved in assessing such risks, including testing methods. Membership in these two independent advisory committees is by invitation from the Chief Medical Officer. In general when dealing with a specific chemical advice is provided from the COM prior to consideration by the COC, in view of the relevance of mutagenicity data when interpreting the available information on carcinogenicity. These committees provided advice on 1,3-butadiene in 1992 (COT, 1992). This followed a
0300-483X/96/$15.00 0 1996 Elsevier Ireland Ltd. All rights reserved PII SO300-483X(96)03448-8
222
R J. FMerl Toxicology 113 (1996) 221-225
comprehensive review of all the available information on the mutagenicity and carcinogenicity of the substance from in vitro, animal and epidemiology data together with considerations of relevant data on toxicokinetics and metabolism. These committees considered all such data (involving over 60 primary publications) so as to obtain as comprehensive as possible a profile of the substance before drawing’conclusions which are provided as a narrative summary. The following conclusions were drawn by the COM: (i) The structure and metabolism of 1,3-butadiene would suggest that it had mutagenic potential; it was metabolised in rodents to epoxides with direct-acting mutagenic potential. (ii) 1,3-Butadiene produced gene mutation in Salmonella typhimurium strains TA1530 and TA1535 in the presence of an exogenous metabolic activation system. Positive results were also reported in the mouse lymphoma assay. No data were available from in vitro metaphase analysis for clastogenic potential, apart from some evidence of SCE induction in CHO cells. (iii) The mutagenic activity of 1,3-butadiene could be expressed in somatic cells in vivo. A clear positive result was obtained in a bone marrow micronucleus test in mice following exposure to 1,3-butadiene by inhalation. There was also some evidence of an increase in SCEs. Negative results were obtained in an in vivo assay to detect UDS in the liver, but the Committee recognised that such an assay was not the optimum method for investigating the mutagenic potential of 1,3-butadiene. There was some evidence of covalent binding to DNA in the liver in a study using a limited protocol. (iv) No data were available on the effects in germ cells, but it would be prudent to assume that 1,3-butadiene had the potential for producing heritable mutations. The conclusions drawn by the COC were as follows: (i) Animal bioassay data indicated that 1,3-butadiene was tumorigenic in both the rat and the mouse. In inhalation studies in the Sprague-
Dawley rat the compound induced follicular thyroid adenomas and Leydig cell adenomas. 1,3Butadiene was a potent multi-tissue carcinogen in the B6C3F, mouse by the inhalation route at concentrations of 20 ppm and above. (ii) Evidence from mortality studies in workers at one butadiene manufacturing site and at several styrene-butadiene rubber sites suggested that exposure to 1,3-butadiene was associated with an increased incidence of malignancies of the haematopoietic/lymphopoietic system. Further data were needed, particularly on exposure levels, before conclusions could be drawn regarding a definite causal relationship. (iii) 1,3-Butadiene should be regarded as a potent genotoxic animal carcinogen and a probable human carcinogen. Since the COM review, new data indicate that butadiene gives positive results in the dominant lethal assay in mice, suggesting mutagenic effects in the male germ cells. The Department of Health is not aware of any other new data published since 1992 which would warrant any significant alterations to the conclusions noted above.
2. Use of mechanistic data to discount positive bioassay results The DH and its advisory committees accept that positive animal bioassay data can be discounted if there are adequate mechanistic data on the induction of tumours in animals to indicate that these are not relevant for humans. Such conclusions have been drawn in the case of dichlorobenzene and other chemicals that induce tumours specifically in the male rat kidney by a similar mechanism, and also for example for phenobarbitone and methylene chloride. All cases to date have involved chemicals that act by a “non-genotoxic” mechanism. These arguments clearly do not apply to 1,3-butadiene since the mono- and di-epoxide pathways occur in humans. It is probable that the compound is a more potent carcinogen in the mouse than humans due to higher levels of the di-epoxide being present,
RJ.
Fielder1 Toxicology 113 (19%)
but concerns cannot be discounted regarding 1,3-butadiene being a genotoxic carcinogen in humans; it is thus prudent to assume the absence of a threshold.
3. Setting air quality standards for genotoxic carcinogens in the UK An approach used in many countries is to define a level of incremental risk of death from cancer considered acceptable by society (e.g. 1 in 10’ of 1 in lo6 lifetime). This involves “sociopolitical” considerations. The level of exposure likely to be associated with this level of risk is then calculated using mathematical models for extrapolating from high dose effects (often from animal bioassay data) to low dose effects expected in humans (e.g. the Linear Multistage Model). In the UK, we have general concerns about the use of such models. We question their lack of validation; their relevance to biological mechanisms; the fact that slopes generated by different models differ greatly and that the resulting estimates of risk may vary enormously for a given compound; and finally that the use of a single point estimate of risk gives an erroneous sense of accuracy. The COC has given general advice on setting standards for carcinogenic air pollutants. Although the prudent assumption of the absence of any safe level was the preferred approach for carcinogens acting by a “genotoxic” mechanism, the need for pragmatism was recognised for compounds that cannot be totally avoided. An approach based on an assessment of the exposure at which the increased risk would not be detectable plus a further safety factor was considered reasonable. A procedure based on the use of expert judgement to estimate the level at which no effects would be expected in the genera1 population has recently been published (Maynard et al., 1995). This is outlined in Fig. 1. This type of approach is used in the UK for deriving air quality standards.
221-225
223
4. UK expert panel on air quality standards
W’AQS) In the UK, the Department of Environment (DOE) has the responsibility for air quality, and they are advised by a DoE/DH Advisory Committee, the Expert Panel on Air Quality Standards (EPAQS). The basic premises used by EPAQS when recommending standards are as follows: A meaningful NOAEL cannot be determined; there is no level of exposure which can be described as acceptable in health terms and in need of no further reduction. There is no case for allowing ambient levels of a genotoxic carcinogen to rise. Although the recommended standard cannot be described as being associated with zero risk it should be associated with what, in practice, is a risk sufficiently low as to be unquantifiable. Progressive reduction in ambient levels is desirable below the recommended standard taking into account costs and feasibility. Standards for compounds with only limited data should not be more relaxed than well understood compound. Standards should be based on a relatively long averaging time, e.g. annual average concentrations.
5. Basis for recommended air quality standard for 1,Ibutadiene The Panel examined the available epidemiology on workers exposed to 1,3-butadiene either in production plants or styrene - butadiene rubber facilities (Meinhardt et al., 1982; Downs et al., 1987; Divine, 1990; Matanoski et al., 1990; Santos-Burgoa et al., 1992; Cowles et al., 1994). It was noted that attempts had been made to categorise exposures using job description data but that these had many problems. Exposure data for people currently exposed to 1,3-butadiene and doing jobs, having the same job description as those identified as being of particular risk from 1,3-butadiene were available. The Panel used the current expose data to obtain a level of exposure for workers identified as prob-
224
R J. Fielder I Toxicology I13 (I 9%) 221-225 CAN A 'NO EXPECTED BE
DIVIDE
BY 10
(WORK
-LIFETIME
(OR DEMONSTRABLE]
lDENTlFlEG
FROU SOUND
HUMAN
EFFECT
LEVEI
EPIDEYIOLOGY
CAN A 'NO EXPECTED EXPOSURE]
EFFECT
LEVEL'
(BIOASSAY
ANIMAL
BE DEFINED DATA]
/
DIVIDE
BY 10
(INTER-SPECIES VARIATION]
CAN COMPOUND BE RANKED AGAINST
INDEX
COMPOUND
1 DIVIDED
I
BY 10
(INTRA-INDIVIDUAL
DIVIDE
VARIATION]
(SERIOUS
DIVIDE
BY 10 EFFECT)
BY 10
(INTRA-INDIVIDUAL VARIATION;
,/ VALUE
IS X GREATER IF YES,
IF NO,
THAN AMBIENT
AIR CONCENTRATION
A = RECOMMENDED X = RECOMMENDED
X
A'
STANDARD STANDARD
Fig. 1. Outline procedure for recommending air quality standards for genotoxic carcinogens.
ably not having been at increased risk of developing these cancers. Consideration of these data led the Panel to conclude that it was unlikely that any excess risk of lymphoma or leukaemia would be detectable at exposures below about 1 ppm. It
was accepted that current exposure levels were likely to be lower than in the past and hence this approach was conservative. The weakness of the exposure data and the uncertainties surrounding this figure were however noted.
RJ. FielderIToxicology 113 (19%) 221-225
The starting point in the derivation of a recommend& standard was the identification of 1000 ppb as the “no detectable effect level” in populations occupationally exposed to 1,3-butadiene. This was divided by 10 to take into account the difference between a working lifetime and a chronological lifetime and by a further 10 to account for intra-individual differences in sensitivity, to give a value of 10 ppb. However in the UK ambient levels of 1,3-butadiene in urban air do not exceed 1 ppb as a running annual average (EPAQS, 1994). The EPAQS (1994) thus agreed to recommend 1 ppb as the air quality standard for 1,3-butadiene as a running annual average.
References COT (1992) Annual Report of the Committeeson Toxicology, Mutagenicity and Carcinogenicity of Chemicals in Food, Consumer Products and the Environment. HMSO, London. Cowles, S.R., Tsai, S.P., Snyder, P.J. and Ross, C.E. (1994) Mortality, morbidity and haematological results from a
225
cohort of long-term workers involved in 1,fbutadiene monomer production. Occup. Environ. Med. 51,323-329. Divine, B.J. (1990) An update on mortality amongst workers at a 1,3-butadiene facility-preliminary results. Environ. Health Perspect. 84 119-128. Downs, T.D., Crane, M.M. and Kim, K.W. (1987) Mortality among workers at a butadiene facility. Am. J. Ind. Med. 12, 311-329. EPAQS (1994) A Recommendation for a UK Air Quality Standard for 1,3-Butadiene. HMSO, London. Matanoski, GM., Santos-Burgoa, C. and Schwartz, L. (1990) Mortality of a cohort of workers in the styrene-butadiene polymer manufacturing industry (1943-82). Environ. Health Perspect. 86, 107- 117. Maynard, R.L., Cameron, K.M., Fielder, R.J., McDonald, A. and Wadge, A. (1995) Setting air quality standards for carcinogens: an alternative to mathematical quantitative risk assessment. Hum. Exp. Toxicol. 14, 175-186. Meinhardt, T.J., Lemen, R.A., Crandall, M.S. and Young, R.J. (1982) Environmental epidemiologic investigations of the styrene-butadiene rubber industry. Stand. J. Work Environ. Health 8, 250-259. Santos-Burgoa, C., Matanoski, GM., Zeger, S. and Schwartz L. (1992) Lymphohematopoietic cancer in styrene-butadiene polymerisation workers. Am. J. Epidemiol. 136, 843-854.
Toxicology 113 (1996) 221-225
Risk assessment of butadiene in ambient air; the approach used in the UK1 R.J. Fielder Department oj Health (HEF Division), Skipton House, Elephant and Castle, London SE1 6L W, UK
The UK Committee on Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (COC) and the related Committee on Mutagenicity provided advice on 1,3-butadiene in 1992. This followed detailed consideration of the available mutagenicity, animal carcinogenicity and epidemiology data plus information on toxicokinetics. They concluded that 1,3-butadiene was an in vivo mutagen, a potent genotoxic animal carcinogen and should be regarded as a probable human carcinogen. The Department of Health is not aware of more recent data warranting reconsideration of these conclusions. General advice on setting air quality standards for carcinogenic air pollutants was given by the COC. Although the prudent assumption of the absence of any safe level for genotoxic carcinogens was preferred, a pragmatic approach based essentially on assessment of the exposure at which no increased risk would be detected, plus a safety factor, was considered reasonable for compounds like butadiene where exposure cannot be totally avoided. This approach, plus recognition that it is unadvisable to allow ambient levels of genotoxic carcinogens to rise, is used in the UK. The procedure by which the Department of Environment’s Expert Panel on Air Quality Standards recommended a value of 1 ppb for butadiene based on these principles is described. Keywords:
1,3-Butadiene; Genotoxic carcinogen; Air quality standard
1. Introduction In the UK, the Department of Health (DH) has established the Committee on Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (COC) and also the Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment (COM). Their charge is to advise Government ‘The opinions presented in this paper are those of the author and in no way commit the Department of Health or other Government Departments.
Departments and Agencies on the carcinogenic or mutagenic risk of chemicals, and on the general principles involved in assessing such risks, including testing methods. Membership in these two independent advisory committees is by invitation from the Chief Medical Officer. In general when dealing with a specific chemical advice is provided from the COM prior to consideration by the COC, in view of the relevance of mutagenicity data when interpreting the available information on carcinogenicity. These committees provided advice on 1,3-butadiene in 1992 (COT, 1992). This followed a
0300-483X/96/$15.00 0 1996 Elsevier Ireland Ltd. All rights reserved PII SO300-483X(96)03448-8
222
R J. FMerl Toxicology 113 (1996) 221-225
comprehensive review of all the available information on the mutagenicity and carcinogenicity of the substance from in vitro, animal and epidemiology data together with considerations of relevant data on toxicokinetics and metabolism. These committees considered all such data (involving over 60 primary publications) so as to obtain as comprehensive as possible a profile of the substance before drawing’conclusions which are provided as a narrative summary. The following conclusions were drawn by the COM: (i) The structure and metabolism of 1,3-butadiene would suggest that it had mutagenic potential; it was metabolised in rodents to epoxides with direct-acting mutagenic potential. (ii) 1,3-Butadiene produced gene mutation in Salmonella typhimurium strains TA1530 and TA1535 in the presence of an exogenous metabolic activation system. Positive results were also reported in the mouse lymphoma assay. No data were available from in vitro metaphase analysis for clastogenic potential, apart from some evidence of SCE induction in CHO cells. (iii) The mutagenic activity of 1,3-butadiene could be expressed in somatic cells in vivo. A clear positive result was obtained in a bone marrow micronucleus test in mice following exposure to 1,3-butadiene by inhalation. There was also some evidence of an increase in SCEs. Negative results were obtained in an in vivo assay to detect UDS in the liver, but the Committee recognised that such an assay was not the optimum method for investigating the mutagenic potential of 1,3-butadiene. There was some evidence of covalent binding to DNA in the liver in a study using a limited protocol. (iv) No data were available on the effects in germ cells, but it would be prudent to assume that 1,3-butadiene had the potential for producing heritable mutations. The conclusions drawn by the COC were as follows: (i) Animal bioassay data indicated that 1,3-butadiene was tumorigenic in both the rat and the mouse. In inhalation studies in the Sprague-
Dawley rat the compound induced follicular thyroid adenomas and Leydig cell adenomas. 1,3Butadiene was a potent multi-tissue carcinogen in the B6C3F, mouse by the inhalation route at concentrations of 20 ppm and above. (ii) Evidence from mortality studies in workers at one butadiene manufacturing site and at several styrene-butadiene rubber sites suggested that exposure to 1,3-butadiene was associated with an increased incidence of malignancies of the haematopoietic/lymphopoietic system. Further data were needed, particularly on exposure levels, before conclusions could be drawn regarding a definite causal relationship. (iii) 1,3-Butadiene should be regarded as a potent genotoxic animal carcinogen and a probable human carcinogen. Since the COM review, new data indicate that butadiene gives positive results in the dominant lethal assay in mice, suggesting mutagenic effects in the male germ cells. The Department of Health is not aware of any other new data published since 1992 which would warrant any significant alterations to the conclusions noted above.
2. Use of mechanistic data to discount positive bioassay results The DH and its advisory committees accept that positive animal bioassay data can be discounted if there are adequate mechanistic data on the induction of tumours in animals to indicate that these are not relevant for humans. Such conclusions have been drawn in the case of dichlorobenzene and other chemicals that induce tumours specifically in the male rat kidney by a similar mechanism, and also for example for phenobarbitone and methylene chloride. All cases to date have involved chemicals that act by a “non-genotoxic” mechanism. These arguments clearly do not apply to 1,3-butadiene since the mono- and di-epoxide pathways occur in humans. It is probable that the compound is a more potent carcinogen in the mouse than humans due to higher levels of the di-epoxide being present,
RJ.
Fielder1 Toxicology 113 (19%)
but concerns cannot be discounted regarding 1,3-butadiene being a genotoxic carcinogen in humans; it is thus prudent to assume the absence of a threshold.
3. Setting air quality standards for genotoxic carcinogens in the UK An approach used in many countries is to define a level of incremental risk of death from cancer considered acceptable by society (e.g. 1 in 10’ of 1 in lo6 lifetime). This involves “sociopolitical” considerations. The level of exposure likely to be associated with this level of risk is then calculated using mathematical models for extrapolating from high dose effects (often from animal bioassay data) to low dose effects expected in humans (e.g. the Linear Multistage Model). In the UK, we have general concerns about the use of such models. We question their lack of validation; their relevance to biological mechanisms; the fact that slopes generated by different models differ greatly and that the resulting estimates of risk may vary enormously for a given compound; and finally that the use of a single point estimate of risk gives an erroneous sense of accuracy. The COC has given general advice on setting standards for carcinogenic air pollutants. Although the prudent assumption of the absence of any safe level was the preferred approach for carcinogens acting by a “genotoxic” mechanism, the need for pragmatism was recognised for compounds that cannot be totally avoided. An approach based on an assessment of the exposure at which the increased risk would not be detectable plus a further safety factor was considered reasonable. A procedure based on the use of expert judgement to estimate the level at which no effects would be expected in the genera1 population has recently been published (Maynard et al., 1995). This is outlined in Fig. 1. This type of approach is used in the UK for deriving air quality standards.
221-225
223
4. UK expert panel on air quality standards
W’AQS) In the UK, the Department of Environment (DOE) has the responsibility for air quality, and they are advised by a DoE/DH Advisory Committee, the Expert Panel on Air Quality Standards (EPAQS). The basic premises used by EPAQS when recommending standards are as follows: A meaningful NOAEL cannot be determined; there is no level of exposure which can be described as acceptable in health terms and in need of no further reduction. There is no case for allowing ambient levels of a genotoxic carcinogen to rise. Although the recommended standard cannot be described as being associated with zero risk it should be associated with what, in practice, is a risk sufficiently low as to be unquantifiable. Progressive reduction in ambient levels is desirable below the recommended standard taking into account costs and feasibility. Standards for compounds with only limited data should not be more relaxed than well understood compound. Standards should be based on a relatively long averaging time, e.g. annual average concentrations.
5. Basis for recommended air quality standard for 1,Ibutadiene The Panel examined the available epidemiology on workers exposed to 1,3-butadiene either in production plants or styrene - butadiene rubber facilities (Meinhardt et al., 1982; Downs et al., 1987; Divine, 1990; Matanoski et al., 1990; Santos-Burgoa et al., 1992; Cowles et al., 1994). It was noted that attempts had been made to categorise exposures using job description data but that these had many problems. Exposure data for people currently exposed to 1,3-butadiene and doing jobs, having the same job description as those identified as being of particular risk from 1,3-butadiene were available. The Panel used the current expose data to obtain a level of exposure for workers identified as prob-
224
R J. Fielder I Toxicology I13 (I 9%) 221-225 CAN A 'NO EXPECTED BE
DIVIDE
BY 10
(WORK
-LIFETIME
(OR DEMONSTRABLE]
lDENTlFlEG
FROU SOUND
HUMAN
EFFECT
LEVEI
EPIDEYIOLOGY
CAN A 'NO EXPECTED EXPOSURE]
EFFECT
LEVEL'
(BIOASSAY
ANIMAL
BE DEFINED DATA]
/
DIVIDE
BY 10
(INTER-SPECIES VARIATION]
CAN COMPOUND BE RANKED AGAINST
INDEX
COMPOUND
1 DIVIDED
I
BY 10
(INTRA-INDIVIDUAL
DIVIDE
VARIATION]
(SERIOUS
DIVIDE
BY 10 EFFECT)
BY 10
(INTRA-INDIVIDUAL VARIATION;
,/ VALUE
IS X GREATER IF YES,
IF NO,
THAN AMBIENT
AIR CONCENTRATION
A = RECOMMENDED X = RECOMMENDED
X
A'
STANDARD STANDARD
Fig. 1. Outline procedure for recommending air quality standards for genotoxic carcinogens.
ably not having been at increased risk of developing these cancers. Consideration of these data led the Panel to conclude that it was unlikely that any excess risk of lymphoma or leukaemia would be detectable at exposures below about 1 ppm. It
was accepted that current exposure levels were likely to be lower than in the past and hence this approach was conservative. The weakness of the exposure data and the uncertainties surrounding this figure were however noted.
RJ. FielderIToxicology 113 (19%) 221-225
The starting point in the derivation of a recommend& standard was the identification of 1000 ppb as the “no detectable effect level” in populations occupationally exposed to 1,3-butadiene. This was divided by 10 to take into account the difference between a working lifetime and a chronological lifetime and by a further 10 to account for intra-individual differences in sensitivity, to give a value of 10 ppb. However in the UK ambient levels of 1,3-butadiene in urban air do not exceed 1 ppb as a running annual average (EPAQS, 1994). The EPAQS (1994) thus agreed to recommend 1 ppb as the air quality standard for 1,3-butadiene as a running annual average.
References COT (1992) Annual Report of the Committeeson Toxicology, Mutagenicity and Carcinogenicity of Chemicals in Food, Consumer Products and the Environment. HMSO, London. Cowles, S.R., Tsai, S.P., Snyder, P.J. and Ross, C.E. (1994) Mortality, morbidity and haematological results from a
225
cohort of long-term workers involved in 1,fbutadiene monomer production. Occup. Environ. Med. 51,323-329. Divine, B.J. (1990) An update on mortality amongst workers at a 1,3-butadiene facility-preliminary results. Environ. Health Perspect. 84 119-128. Downs, T.D., Crane, M.M. and Kim, K.W. (1987) Mortality among workers at a butadiene facility. Am. J. Ind. Med. 12, 311-329. EPAQS (1994) A Recommendation for a UK Air Quality Standard for 1,3-Butadiene. HMSO, London. Matanoski, GM., Santos-Burgoa, C. and Schwartz, L. (1990) Mortality of a cohort of workers in the styrene-butadiene polymer manufacturing industry (1943-82). Environ. Health Perspect. 86, 107- 117. Maynard, R.L., Cameron, K.M., Fielder, R.J., McDonald, A. and Wadge, A. (1995) Setting air quality standards for carcinogens: an alternative to mathematical quantitative risk assessment. Hum. Exp. Toxicol. 14, 175-186. Meinhardt, T.J., Lemen, R.A., Crandall, M.S. and Young, R.J. (1982) Environmental epidemiologic investigations of the styrene-butadiene rubber industry. Stand. J. Work Environ. Health 8, 250-259. Santos-Burgoa, C., Matanoski, GM., Zeger, S. and Schwartz L. (1992) Lymphohematopoietic cancer in styrene-butadiene polymerisation workers. Am. J. Epidemiol. 136, 843-854.