Document dYaYBveGgLg1b0pB5nQydnLY9

r ' - Research Articles Experientia 48 (1992). Birkhauser Verlag, CH-4010 Basel/Switzerland 301 Correlation between car ownership and leukaemia: Is non-occupational exposure to benzene from petrol and motor vehicle exhaust a causative factor in leukaemia and lymphoma? s.P. Wolff Toxicology Laboratory, Department of Clinical Pharmacology, University College and Middlesex School of Medicine, 5 , University Street, London WClE 6JJ (England) Received 24 May 1991; accepted 12 September 1991 Abstract. Although there is widespread agreement that many cancers have environmental causes we are often unable to see associations between specific cancers and exposure to environmental chemicals, One might also speculate that the more widespread, common-place and `normal' a chemical exposure is perceived to be then the less likely it will be that the exposure is recognised, let alone be considered to cause cancer. Widespread contamination of air by chemicals associated with internal combustion may be an example of one such `invisible' carcinogenic exposure. Yet evidence is available which suggests that many leukaemia and lymphoma cases, as well as other cancers, may be caused by this mundane and ubiquitous environmental contamination. The hypothesis is developed that leukaemia `clustering'as well as national leukaemia incidence may be related to non-occupational exposure to benzene formed I by petrol combustion and resulting from petrol evaporation The possible association between exposure to fuel vapours, internal combustion products and cancer merits much closer examination than it receives at present. Key words. Leukaemia; benzene; motor vehicle; air pollution; socioeconomic indicators. "Almost all cancers appear to be caused by exposure to dealkylation of alkylbenzenes during internal combus- i factors in the environment. The most promising approach tion 12- .14 Levels of airborne benzene are closely related to the control of the diseuse is to identifv thosefactors and to traffic density and, although generally in the range I eliminate them"2. of 1-4parts per billion (ppb) in urban areas, have been reported to reach 150ppb in densely-trafficked areas 16. ! "Epidemiological evidence suggests that present levels of Highest levels of benzene are, however, present within chemical use do not lead 10 widespread harmful contamina- the cars themselves and may reach US occupational ex- tion of the human en~ironment"~. posure limits (780ppb compared with 1000ppb) al- \ though the typical range is in the order of 10-20ppb". I Exposures during vehicle refuelling are, briefly, in the Locally increased rates of leukaemia in the vicinity of range of a few parts per million nuclear plants such as the Sellafield nuclear reprocessing The hypothesis, then, is that non-occupational exposure ! plant in Cumbria, UK, have been noted and have to benzene from petrol evaporation and car exhaust con- aroused considerable interest`. The possibility that fac- tributes to national leukaemia incidence. Specifically, the I tors other than ionizing radiation could contribute to the higher levels of benzene exposure encountered by travel- I elevated incidence of leukaemia/lymphoma in the vicini- ling inside cars, or being in cars during refuelling would ty of such installations has been raised by 1) the observa- add substantially to the risk accumulated as a result of tion that elevated local rates of leukaemia occur at sites long-term exposure to ambient levels of airborne benzene distant from nuclear power installations and by 2) re- and would thus contribute to increased incidence of ports that area incidence of lymphoproliferative cancer leukaemia in regions where a greater use of cars is pre- increaseswith higher area socioeconomicstatus 6-11.Ar- dicted to occur. eas close to nuclear installations appear to have a high Proportion of higher occupational class households so Methods that the risk may simply be associated with higher socio- The Leukaemia Research Fund has mapped the inci- economic status, rather than radiation. dence of leukaemia and lymphoma in 22 counties in the Since car ownership is not homogeneous throughout SO- UK for the years 1984-198819. Data of relative risk by b CietY (but is concentrated within the higher socioeconom- county from the Atlas can be related to information on strata) it is conceivable that the leukaemogenic risk car ownership by county for a slightly earlier period factor associated with a higher socioeconomicstatus is a (1981) using the Great Britain Population Census2' and Potentially greater level of exposure to chemical leukae- published Transport Statistics21. mogens associated with motor vehicle usage. Petrol, diesel fuel and internal combustion emissions contain Results and discussion Potentially carcinogenic hydrocarbons, such as The table shows Spearman rank correlation coefficients which is a known human leukaemogen. between incidence of leukaemia/lymphoma (age- and Benzene is present at a concentration of circa 5 % v/v in sex-standardized relative risks for 22 counties and 21 072 Europeanpetrol, and is also produced by the thermolytic total cases in England and Wales from 1984-1988) and 302 Experientia 48 (1992), Birkhauser Verlag, CH-4010 Basel/Switzerland Research Articles Spearman rank correlation coeflicients ~ ~~ Diagnostic group Number of cases Cars per household Cars per thousand All myeloproliferative disease 6222 0.17 0.12 Acute myeloid leukaemia Chronic myeloid leukaemia Myeloid dysplastic states Other myeloproliferative disease 1991 0.38* 678 0.08 1806 0.21 1747 0.07 0.32 0.20 0.10 0.07 All lymphoproliferative disease 14850 Acute lymphoblastic leukaemia Chronic lymphoid leukaemia Low-grade non-Hodgkin lymphoma High-grade non-Hodgkin lymphoma Hodgkin disease Multiple myeloma * p < 0.05; **p < 0.025; ***p < 0.001 775 3340 2415 2550 1799 3223 0.40* OM** 0.41 * 0.66*** 0.21 0.21 0.27 0.35 0.35 0.34 0.68 *** 0.08 0.10 0.18 car ownership. There is a statistically significant association between car ownershipand acute myeloid leukaemia (which comprises 9 `30of total cases in the Leukaemia and Lymphoma Atlas), the cancer specifically associated with benzene exposure, as well as between all lymphoproliferative diseases (70 % of total Atlas leukaemia/lymphoma cases) and car ownership. Within the category of lymphoproliferative disease there is greatest association with car ownership for low grade non-Hodgkin's lymphoma (comprising 11YO of total leukaemia/ lymphoma cases reported in the Atlas) but a significant association also exists for acute lymphoblasticleukaemia (4 % of total cases), the type common in children, as well as for chronic lymphoid leukaemia (16% of total cases). Associations appear generally stronger (achieving statistical sigmfkance in most cases) when calculated against cars per household rather than cars per thousand population. This could be taken to indicate association of personal risk of leukaemia with car ownership rather than association of risk simply with local car density. Figure 1shows data distribution for all lymphoproliferative disease as well as low-grade non-Hodgkin lymphoma and car ownership, confirming the geographical association. The higher the county level of car ownership, the greater the risk of lymphoproliferative disease. Greater levels of car ownership could, of course, merely reflect county differences in ruralturban status and thus the presence of some leukaemogenic factor present in rural areas to a greater extent. However, correlation coemcients remain essentially unchanged when corrected for population density (data not shown) so that differences in rural/urban status appear not to be an important factor in the correlations observed. Correlations of home ownership and overcrowding with leukaemia were, by contrast, close to zero 22. These cross-sectional epidemiological data based on geographical differences in cancer incidence and car owners h p must, of course, be regarded as preliminary and cannot show that car ownership is a causative factor in leukaemia/lymphoma. But if the associations are real LOW GRADE NON HODGKIN LYMPHOMA 160150- 0 140130- 0. 3 80- 70- 6050 i 0.5 0 0.6 0.7 0.8 0.9 CARS PER HOUSEHOLD 1.o 1301 ALL LYMPHOPROLIFERATIVE DISEASE 0 ?vr): ,201 iz 110- 2w 100- eI 3 90- 2 0. a. 0 0 0.5 0.6 0.7 0.8 0.9 1.o CARS PER HOUSEHOLD Figure 1. Data distribution for relationship between low-grade nonHodgkin lymphoma, all lymphoproliferative disease and level of car ownership. then it is probable that benzene, a well-characterised leukaemogen, is responsible since it causes leukaemia and can reach high levels in cars13.14.The correlations between car ownership and leukaemia are also in agreement with those obtained by Robinson 23 who showed a strong relationship between Australian leukaemia mortality and vehicle usage, as monitored by the rate of vehicle fatalities. A greater level of car ownership and usage at certain locations compared with the national or regional average, could thus (according to this hypothesis) be responsible for the emergence and detection of leukaemia `hot spots' as small, but statistically significant fluctuations above leukaemia incidence baseline, if non-occupational levels of benzene can cause cancer. A corollary to the hypothesis that a greater local level of car ownership/ usage is responsible for a locally-elevated incidence of leukaemia, would be that national and regional background leukaemia incidence is related to the same underlying causative factor. This would be an attractive explanation for the large increase in childhood and adult leukaemia from the early 1920's to the end of the 1 9 6 0 ' ~ ~ T~h*e~`c'l.usters' might thus be the visible tips of large icebergs of benzene-related leukaemia. - Research Articles Expenentia 48 (1992). Birkhauser Verlap. CH-4010 BaseliSwrtzerland 303 There is little doubt that exposure to the high levels of benzene associated with many industrial settings causes leukaemia 26. ". On the basis of epidemiologyperformed on exposed worker populations it has been estimated that there will be 10 excess deaths per 1000 employees exposed for a working lifetime to 1 part per million, the current US occupational exposure standard 2 8 . This cor- responds to a unit risk (estimated by the US Environ- mental Protection Agency2') of 7 x for death due to leukaemia as a result of lifetime exposure to 0.3 ppb (1 pg/m3).The World Health Organisation sets the unit risk somewhat lower at 4 x 'O. Both these risk esti- mates assume no `threshold' and a linear exposure-can- cer dose-response curve. Acute myeloid leukaemia is the most common form of leukaemia in benzene-exposed adults (for which the above unit risk estimates have been calculated) but acute lymphoblastic leukaemia, chronic myeloid leukaemia, lymphoma and multiple myeloma may also occur 31 -36. There is evidence that car-derived carcinogen exposure might cause leukaemia since Savitz and Feingold ob- served that rates of leukaemia were higher in areas of higher traffic density in a study of childhood leukaemia incidence in Denver, Colorado 37. The authors, although cautious about relating their observations to benzene, found that children living in areas with more than 10,000 vehicles a day had an odds ratio (OR) of 4.7 (confidence interval 1.6-13.5) for leukaemia. The OR was 2.7 for more than 5000 vehicles per day and was greater for 0-4 year olds (OR = 5.6) in the `exposed' group; defined as those living in areas with more than 500 vehicles per day. Consistent with a hypothesis of motor vehicle emissions as risk factor for cancer it has been shown that rates of leukaemia and other cancers are higher in areas of higher traffic density and correlate with local deposition of vehi- cle-derived hydrocarbons 38 -41. The hypothesis that low levels of environmental benzene exposure associated with fuel vapours and internal com- bustion could be responsible for leukaemia may be greet- ed with scepticism since there remains a residue of debate about the carcinogenic potential of exposure to benzene even in the 1-10 ppm range42. Certainly, although very high (occupational) levels of benzene are linked with leukaemia l4 any risk of cancer at lower levels of benzene appears to disappear into the background of `sponta- neous' leukaemia incidence (fig. 2). The existence of the background rate, and the shape of the graph in figure 2, is used as an important justification for the concept of the `threshold` dose which, translated into biological terms, argues for the presence of detoxification and DNA re- Pairhaintenance mechanisms which operate efficiently at lower concentrations of carcinogen but which can be saturated at higher doses. However, such an interpreta- tion depends very much on how one views this graph (fig. 3). The hypothesis that the background `sponta- neous' leukaemia incidence may also be related to non- Occupational benzene exposure casts the threshold dose t RESPONSE BACKGROUND INCIDENCE GENERAL OCCUPATIONALLY POPULATION EXPOSED -DOSE Figure 2. Dose-response curve for relationship between benzene exposure and leukaemia assuming that background incidence of leukaemia is `spontaneous' and unrelated to non-occupationalexposure. I THRESHOLD UNKNOWN RESPONSE DOSE INCIDENCE -DOSE Figure 3. Revised dose-response curve for relationship between benzene exposure and leukaemia suggesting that background incidence is related to non-occupational benzene exposure and that threshold dose is arbitrary dividing line between known and unknown doses. in a new light. In this instance, the threshold becomes the artificial dividing line between the high (and fairly welldefined) levels of occupational exposure, to which thousands of adults are exposed, and the less-defined and generally lower levels to which millions of people are exposed in the course of their day-to-day activities. This background level of exposure to benzene is suggested to contribute to the background level of leukaemia and lymphoma in the non-occupationally exposed population. The Leukaemia and Lymphoma Atlas study population was 15,265,000individuals (both sexes, all ages) of which 1991 were diagnosed as having acute myeloid leukaemia in the 5-year period: i.e. an annual risk of death from acute myeloid leukaemia of circa 3 x lo-'. Assuming that diagnosed individuals had already received their `lifetime' dose this would correspond to an exposure of 1-2 ppb (70-140 ppb-years), using the risk estimates given above, and thus well in the range of the non-occupational exposure anticipated to occur as a result of car usage. Various large-scale surveys of risk factors for leukaemia and lymphoma are now being planned by Government agencies and charities concerned with leukaemia43. There is considerable interest in the analysis of factors such as radon exposure and unusual responses to infec- 304 Experientia 48 (1992),Birkhauser Verlag, CH-4010 Basel/Switzerland Research Articles I tions. However, analysis of a dominant source of leukaemogen exposure, namely non-occupational exposure to benzene as a result of vehicle usage and traffic, is not being considered. Clearly, it is difficult to assess differences in levels of an exposure which is now virtually ubiquitous in developed countries and considered a healthy and desirable aspect of national development. Indeed, Lutz and colleagues have recently reviewed the evidence which suggests that regulation of benzene exposure is controlled more by psychology, sociology, economics and politics rather than medical or scientific con- 16 Thorburn, S.,and Colenutt, B. A., Int. J. envir. Studies 13 (1979)265. 1 i:17 Gilks, J. M. L., Boelsma-v. Houte, E., Eyres, A. R., Rousseaq G Roythorne, C., Stanton, D. W., Thomas, F. C., and Simpson, B. Health risk of exposure to non-occupationalsources of bemew, in: CONCAWE (the oil companies' European organisation for environmental and health protection) report NO. 8/89.The Hague, Netherlands 1989. I I I I 18 Brief, R. S.,Lynch, J., Bemath, T., and Scala, R. A., Am. ind. H ~ ~ . ASS. J. 41 (1980)616-623. I 19 Cartwright, R.A., Alexander, F.E., McKinney, P. A., and Rickets, T. J., Leukaemia and Lymphoma: an atlas of distributionwithin areaS 1 of England and Wales 1984-1988. London, Leukaemia Research Fund, 1990. I 20 Oftice of Population Censuses and Surveys. Census 1981,Key Statis- tics for Local Authorities, HMSO, 1984. siderations 44. Detailed measurements of benzene and hydrocarbon exposure would, however, allow support for, or refutation of the hypothesis that internal combustion-powered vehicles cause cancer. There is a pressing need for biological and epidemiological investigation of non-occupational benzene exposure. 21 Department of Transport, Transport Statistics Great Britain 1971- 1981. HMSO, 1982. 22 Wolff, S.P., Leukaemia and Radon. Nature, 352 (1991)288. 23 Robinson, A. A., Med. Hypoth. 9 (1982)345. 24 Court-Brown, W. M., and Doll, R., Br. med. J. April (1961)981. 25 Davis, D.L.,Hwl, D., Fox, J., and Lopez, A., Lancet 336 (19%) 474. 26 Goldstein, B. D., Occup. Med.: State Art Rev. 3 (1988)541. 27 Brandt, L.,Nilsson, P. G., and Mitelman, E, Br. Med. J. I(1978)553. I I ! I 28 Occupational Exposure to Benzene: Final Rule (29CFR Part 1910), Acknowledgments. I am grateful to John Adams and George Stern for Federal Register 52:176 (11th Sept 1987)pp 34459-34578. helpful discussions. 29 Goldstein, B. D.,Risk assessment and risk management of benzene I by the environmental protection agency. Banbury Report f9 (1985) 1 Wolff, S. P., Child Leukaemia - Curies or Cars? Nature 346 (1990) 293-304. 517. 30 World Health Organisation: Air quality guidelines for Europe. WHO 2 Cairns, J., The Cancer Problem. Sci. Am., Nov. (1975)64. Regional Publications, European Series No. 23, 1987. I 3 McLean, A. E. M., Hazards from chemicals: scientific questions and conflicts of interest. Proc. R. SOC.Lond. B. 205 (1979)179-197. 4 Black, D., Investigation of the possible increased incidence of cancer in West Cumbria. HMSO Publications, London 1984. 31 Aksoy, M., Erdem,S.,and Dincol, G., Acta haemat. 55 (1976)65. 32 Enterline, P.E., Lancet 2 (1979)1021. 33 Vianna, N. J., and Polan, A., Lancet I(1979) 1394. 34 Rinsky, R. A., Smith, A. B., Hornung, R., Filloon, T. G., Young, I I 5 Kinlen, L. J., Clarke, K., and Hudson, C., Evidence from population mixing in British New Towns 1946-85 of an infective basis for child- hood leukaemia. Lancet 336 (1990)577. 6 Peto, J., Radon and the risks of cancer. Nature 345 (1990)389. R. J., Okun, A. H., and Landrigan, P. J., N. Engl. J. Med. 316 (1987) 1044. 35 Infante, P. E, and White, M. C., Am. J. ind. Med. 7 (1985)403. 36 Young, N., Am. J. ind. Med. 15 (1989)495. 7 Cook-Mozaffari, P. J., Darby, S. C., Doll, R., Fonnan, D., Hermon, 37 Savitz, D. A., and Feingold, L., h n d . J. Work envir. Health f5 C., Pike, M. C., and Vincent, T., Br. J. Cancer 59 (1989)476. (1989)360. 8 Alexander, F.E., Rickets, T. J., McKinney, P. A., and Cartwright, R. A., Lancet 336 (1990)1461. 9 Sanders, B. M., White, G. C., and Draper, G. J., J. Epidemiol. C o r n . Health 35 (1981)245. 38 Blumer, W., Jaumann, R., and Reich, Th., Schweiz. med. Rundschau (PRAXIS) 61 (1972)514. 39 Ippen, M., Fehr, R.,and Krasemann, E. O., Versicherungsmediin2 (1989)39. I ! 10 McWhirter, W. R., Br. J. Cancer 46 (1982)640. 11 F a d , E., Jackson, E. W., and Klauber, M. R., Birth characteristics and leukaemia in childhood. J. natl Cancer Inst. 47 (1971)509. 12 Carey, P. M., US Environmental Protection Agency TechnicalReport 40 Blumer, W., Blumer, M., and Reich, Th., Fortschr. Med. 95 (1977) 1497 and 1551. 41 Blumer, M., Blumer, W., and Reicht, Th., Env. Sci. Tech. I1 (1977) 1082. I I EPA-AA-TSS-PA-86-5, 1987. 42 Graham, D., Green, L. C., and Roberts, M. J., In Search of Safety: 13 InternationalAgency for Research on Cancer. IARC Monographson Chemicals and Cancer Risk. Harvard University Press, 1988. I the evaluation of carcinogenic risks to humans, vol. 46.Diesels and Gasoline Engine Exhausts and Some Nitroarenes. IARC, Lyon, France 1989. 14 InternationalAgency for Researchon Cancer. IARC Monographson 43 Aldhous, P., Nature 347 (1990)508. 44 Lutz, W. K., Poetzsch, J., Schlatter, J., and Schlatter, C., Trends phar- mac. Sci. 12 (1991)214. I I the evaluation of carcinogenic risks to humans Volume 29. Some !' Industrial Chemicals and Dyestuffs, pp 95-148 IARC, Lyon, France 1982. +15 World Health Organization:Air quality guidelines for Europe. WHO 0014-4754/92/030301-04S~O1 0.20/0 regional publications. European Series No. 23, 1987. 0Birkhluser Verlag Basel, 1992 I I II I I I I I I I I I i;