The Effects of Polycyclic Aromatic Hydrocarbons on Human Health

14. It has already been noted that humans are always exposed to environmental PAHs in the form of mixtures and not as individual compounds, a fundamental point which is discussed later in more detail. Men exposed occupationally to high concentrations of PAH mixtures show an increased incidence of tumours of the lung, skin and possibly bladder and other sites. Of these various tumours, lung cancer is the one most obviously linked to exposure to PAHs through inhaled air. In addition, several of the individual substances present in PAH mixtures have been shown to be carcinogenic in experimental animals when tested as single purified compounds.

15. PAHs are present in ambient air as gases or adsorbed onto particles, both of which can be inhaled into the lungs. Most carcinogenic PAHs occur almost exclusively in the long-lived particulate phase. Irrespective of their physical properties and initial routes of exposure, the tumour-inducing effects of carcinogenic PAHs depend on their conversion in the body into compounds which, unlike the parent substances, are capable of reacting with and damaging the genetic material (DNA) in the nuclei of cells. Such compounds are defined as genotoxic carcinogens. The activation process is mediated by various enzyme systems in susceptible target tissues. It is known that PAHs are rapidly absorbed and can be activated in the lung (IARC, 1983).

16. Individual PAHs vary in their carcinogenic activity in experimental animals and, for several compounds, their capacity to induce tumours is uncertain or unknown. Some are more fully documented than others and information for B[a]P, in particular, is extensive. In 1983 and 1987, the International Agency for Research on Cancer (IARC) classified 48 PAHs according to their likely human carcinogenicity. IARC's evaluation of 16 PAHs commonly found in ambient air is summarised below. The three potent animal carcinogens benzo[a]pyrene, benz[a]anthracene and dibenz[ah]anthracene are classified as 'probably carcinogenic to humans'. The qualification ‘probably' reflects the absence of any data on the risks associated with human exposure to these compounds in isolation.

17. The implications of human exposures to mixtures of PAHs, rather than to individual substances, are important. Such mixtures are complex, incompletely characterised in terms of all their PAH constituents, and likely to vary in composition over the long periods of time during which PAH-induced tumours develop. More theoretically, the possible modes of interaction between different PAHs in such mixtures are not fully understood. Various approaches have been adopted to quantitate exposure to these mixtures, such as measuring benzene soluble material, total PAH levels, or the levels of one specific compound - usually benzo[a]pyrene - which is often regarded as a representative ‘marker' substance.

18. In considering the evidence of human exposure to PAHs and their effects on health, the Panel has taken account of two types of study. In one type, of workers exposed to PAHs in industrial situations, useful information on exposure is sometimes available but the concentrations are many times greater than those encountered in the general environment. The second type of study, of large populations of people living in cities, has the advantage of being more relevant to the effects of ambient air, but such studies rarely contain information on the concentration and/or duration of exposure of the population. In the following paragraphs we summarise the evidence that we have obtained from both occupational and environmental studies.

Occupational studies

19. The industrial epidemiology of exposures to PAHs in air is based mainly on workers employed in gas and coke production and in aluminium refining. Two preliminary points should be emphasised. First, tobacco smoking, which is by far the most important single cause of lung cancer and an important source of PAHs in its own right, is a major confounding factor. Industrial workforces have often contained a high proportion of smokers, and allowance for this has to be made when evaluating the influence of any workplace exposures on lung cancer risks. Secondly, some studies of workforces exposed to PAHs have also shown an excess of bladder cancer, for example among individuals employed in smelting aluminium. These workers may well, however, have been exposed to known bladder carcinogens (notably certain aromatic amines) as well as to PAHs. Information concerning the role of PAHs in the causation of bladder cancer is limited (compared to lung cancer) and difficult to evaluate, and the Panel has concentrated on lung cancer as the most relevant outcome for PAH exposure from the air.

20. The production of gas and coke from coal releases fumes containing PAHs into the workplace, and many investigations of workers in these industries have demonstrated significantly increased risks of lung cancer. In general, the excess risks of lung cancer have been shown in the workers with the highest and most prolonged exposures. For example, in a British study of coal gasification workers in the 1960s (Doll et al., 1965; Doll et al., 1972), workers with heavy exposure in the carbonising plants showed an excess risk of lung cancer, while workers with intermittent exposure or exposure elsewhere in the workplace had no comparable excess. Workers exposed to high concentrations had an approximately 80% increase in the risk of lung cancer over and above other workers. A more recent investigation of coke oven workers in the USA demonstrated a significant excess of deaths from respiratory cancer, with the greatest increase in risk again occurring in those workers with the highest exposure to coke oven emissions (Costantino et al., 1995).

21. Some indications of the atmospheric PAH levels to which workers may have been exposed come from measurements made in studies of coal gasification plants in the UK in the 1960s. Long-term concentrations (averaged over several weeks) of B[a]P were around 3µg/m³**, with peak values above retorts of over 200µg/m³ (Lawther et al., 1965). Similar results were found in gas works in Sweden (Lindstedt and Sollenberg, 1982). Studies in coking plants in the USA, Scandinavia and Czechoslovakia have shown concentrations of B[a]P of 10-40µg/m³ on oven tops and some 10 to 20 fold lower in the general atmosphere of the workplace (Jackson et al., 1974; Lindstedt and Sollenberg, 1982; Masek, 1971).

22. Epidemiological investigations of workers in aluminium refineries in the USA, Canada and Norway have shown significant excesses of deaths from lung cancer related to prolonged exposure to PAH fumes (Ronneberg and Andersen, 1995; Milham, 1979; Gibbs, 1985; Armstrong et al., 1994; Andersen et al., 1982). The Panel considered that a study by Armstrong and others of lung cancer deaths in a sample of men who had worked for at least one year in an aluminium smelter in Canada was particularly helpful as it addressed the problem of smoking by exposed workers (Armstrong et al., 1994). In this investigation exposure to B[a]P as a marker of PAH exposure was estimated for workers in each type of job within the plant. The heaviest exposure occurred for workers in two parts of the process known as ‘the pot room' and ‘anode manufacture', where B[a]P concentrations were 20-40µg/m³. After adjustment for confounding by cigarette smoking and age, a clear association was found between increased exposure to B[a]P and lung cancer deaths (see Table 4).

Table 4. Lung cancer mortality in aluminium production workers in Arvida, Quebec, Canada, 1950-1988 (from Armstrong et al., 1994).

Notes

^a Cumulative exposure to total (particulate and gaseous) benzo[a]pyrene in micrograms per cubic metre of air multiplied by the number of years of exposure.

^b The percentage increase in risk of death from lung cancer compared to those with exposure <10µg/m³ years (the control group).

^c Technically the 95% Confidence Interval.

23. A comparison of the various occupational studies including coal gasification workers, coke oven workers and aluminium smelter workers has revealed that estimates of the increase in risk of lung cancer for those occupationally exposed to a given level of PAHs fall within a relatively narrow range; estimated levels of risk associated with exposure to PAHs in these different occupational environments are similar (Gibbs, 1997).

Environmental studies

24. Epidemiological studies of exposures to levels of PAHs commonly encountered in the general environment have more direct relevance to risks to the health of the public than occupational studies which involve higher exposures. However, such studies of the general environment suffer from a lack of direct measurement of PAHs. A review published in the US of epidemiological evidence on ambient air pollution and lung cancer has suggested that combustion-source air pollution contributes to the occurrence of lung cancer among the general population (Cohen and Pope, 1995). Quantification of the excess lung cancer risk associated with air pollution was limited due to errors in the measurement of exposure levels and confounding by other risk factors such as cigarette smoking. However, the excess risk due to air pollution was considered to be small in comparison to the risk associated with cigarette smoking.

25. In a paper published in 1981, Doll and Peto reviewed studies which extrapolated from occupational data to estimate the lung cancer risk associated with air pollution exposure in urban environments (Doll and Peto, 1981). It was estimated that carcinogenic air pollutants, such as B[a]P, may at one time have contributed to around 10% of lung cancers in urban areas. However, by the late 1970's levels of air pollution had fallen considerably, so that exposures to B[a]P and associated combustion products which occurred at that time were unlikely to account for more than 1% of cases of lung cancer. It is likely that the development of lung cancer reflects exposure over decades, the heavier exposures that occurred more than 20 years ago having played a dominant part. In the time since publication of Doll and Peto's review, air pollution levels have fallen even further, and so it may be assumed that the future risk of lung cancer which is attributable to current levels of air pollution has been further reduced.

26. A study of lung cancer and air pollution in Denmark in the 1970s and 1980s has shown that the risk of this disease in urban dwellers living in Copenhagen was roughly double that of those living in rural areas of the country (Engholm et al., 1996). The authors estimated that B[a]P concentrations in Copenhagen were up to 10 ng/m³ at the relevant time. However, most of the excess lung cancer risk was attributable to smoking and occupational exposure, and a separate effect of air pollution was not demonstrable.

27 By contrast, a study of lung cancer in professional drivers in Sweden showed an increased risk among those working in the city of Stockholm (Jakobsseen et al., 1997), especially in short-distance lorry drivers; these findings allowed for the effects of smoking and the authors suggested that exposure to traffic fumes might have been responsible, although no data were available on atmospheric concentrations of PAHs within the cabs. A study in Denmark, which focused upon employed men who were diagnosed with lung cancer between 1970 and 1989, also revealed a 31 to 64% excess risk of lung cancer in bus, lorry and taxi drivers (Hansen et al., 1998). The smoking habits of Danish drivers and those in other types of employment were found to be similar. However, in the absence of exposure measurements, the contribution of PAH exposure to this result is unclear.

Summary

28. There is clear evidence that PAH mixtures are carcinogenic in humans and several individual PAHs are carcinogenic in experimental animals. Cancers result from exposure to PAHs over several decades. Increased risks of lung cancer, in particular, have been associated with increased concentrations of PAHs in the workplace. There is epidemiological evidence of an association between urban air pollution and excess risk of lung cancer in those most highly exposed, but risks cannot be estimated confidently from these data because of possible confounding by smoking and lack of information on historical exposure levels.

** 1µg/m³ is one millionth of a gram in every cubic metre of air.