Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
that BaP is not a good surrogate for PAHs in mixtures from different sources, although more information is available on its effects than those of other PAHs. However, a person who lives where the air contains BaP at 10 ng/m3 and who breathes 15 In of air per day would breathe in roughly the same amount of BaP as he would from smoking five old-style cigarettes (as discussed by Hoffmann et al. ). It is therefore not unreasonable to assume that this degree of pollution, whicl' was very common only 20 yr ago, may cause a s ignificant amount of lung cancer. Studying the problem directly proves difficult, because one must he especially careful to ensure that an observed effect is not attributable to differences in smoking habits between high- and low-pollution areas. The lung-cancer incidence is affected not only by the number of cigarettes smoked, but by the tar content of the cigarettes, by how far down the cigarette is smoked, and by smokers' ages at starting to smoke and at stopping (if ever); all these aspects of smoking habits have to be considered. It is impossible to allow for all these factors accurately, so extrapolating from an extreme situation, in which small smoking-habit differences can be ignored, is likely to be the best method of estimating general air-pollution effects. Men employed in some occupations are exposed intermittent ly to BaP in air at up to 16,000 ng/m ,1 and they provide an opportunity to study lung-cancer effects in an extreme situation. OCCUPATIONAL EXPOSURE - Many epidemiologic studies of lung cancer have involved occupa- tional exposure to PAH-containing mixtures.2633~ They showed that exposure to high concentrations of PAH-containing mixtures increases the risk of lung cancer. Assuming that the exposed and nonexposed workers have the same smoking habits and that their observed lung-cancer incidence rates are r and rn, respectively, we can express the lung-cancer burden from the exposure either as a ratio, R = re/rn, or as a difference, D = re ~ rn. For general risk-assessment purposes, we can express - these on the basis of per-unit exposure by divid ing R or D by the "expo sure dose . " Both R and D are valid measures of the risk to the occupational group as a group, but they implicitly make very different assumptions about the risks to individual members of the group with different smoking habits. The relative-risk index, R. implicitly assumes that the risk of lung cancer is increased in proportion to the individual's "underlying" risk--a nonsmoker's risk is multiplied by R. and a 2-packs/d smoker's risk is also multiplied by R. The additional risk of the 2-packs/d smoker is thus an order of magnitude greater than the additional risk of the nonsmoker and double the risk of a 1-pack/d smoker. This multiplicative (sometimes referred to as synergistic) phenomenon appears to hold for lung cancer caused "Jointly" by asbestos exposure and cigarette-smoking.23 C-S
The additional-risk index, D, implicitly assumes that the amount of increased risk of lung cancer is independent of other lung-cancer risk--a nonsmoker's risk is increased by the same absolute amount as a 2-packs/d smoker's risk. None of the occupational studies of exposure to PAH-containing mixtures and lung cancer was conducted in such a way as to provide data to help in distinguishing between the possible models (i.e., multiplicative, additive, or somet'ning intermediate). Studies comparing urban and rural lung-cancer rates (or rates in "heavily polluted" and "lightly polluted" areas) in persons with different smoking habits do provide relevant data, but the studies generally have few deaths and go not clearly identify the correct model. Data from the study of Stocks3 (Table C-2) and from the study of Hitosugil6 (Table C-3) illustrate the point. In both studies, the data from the smokers are in good agreement with an additive model for the effect of "air pollution," and these data provide no evidence for a multiplicative model. The data from nonsmokers, however, confuse the picture. In Stocks's study, the effect of air pollution is smaller in the nonsmokers; in Hitosugi's study, there is no effect in nonsmokers. The problem may simply result from basing the rates on such small numbers of deaths in nonsmokers or from misclassifying the smoking habits of a few persons who died of lung cancer. Because the additive model provides such a good fit to the data on smokers, we have assumed this model in our discussion of lung-cancer risk from occupational exposure to PAH-contain · @ long mixtures. To use occupational studies for risk-assessment purposes, we must assume that, as far as lung cancer is concerned, occupational exposures can be expressed as cigarette equivalents, i.e., that the form of Equations 1 through 4 will hold for the excess lung cancer from such exposure. We saw when discussing lung cancer and cigarette-smoking that dose and duration of exposure are critical in determining lung-cancer risk. The occupational studies must, at a minimum, provide a quanti- tative estimate of the dose and duration of exposure to PAH-containing mixtures. With this information and comparative information on the smoking habits of the exposed and nonexposed workers, we can estimate the absolute risk from such exposure. Unfortunately, only one occupational study with high exposure to a PAH-containing mixture supplied even this minimal information.l° UNITED KINGDOM GASWORKERS . . In a prospective study, Doll et al.6'L0 followed a cohort of carbonization workers in British gasworks for up to 12 yr. Carboni- zation workers were exposed to BaP at an estimated average air concen- tration of 3, 000 ng/m3 during an 8-h shift20 and experienced a 142% increase in lung-cancer mortality, compared with their nonexposed workmates (Table C-4~. Although the smoking habits of only some 10t of the cohort were ascertained by Doll and his colleagues,6 the exposed and C-6
nonexposed workers appear to have had very similar smoking habits, with ^' average current consumption of approximately 10 cigarettes/d. It is reasonable, therefore, to assign the excess lung cancer in the exposed group to their working conditions, specifically to the air to which they were exposed. The current age of a smoker and his age at,starting to smoke are both important in determining his risk of lung cancer. Likewise, both current age and age at starting as a carbonization worker are important in deter- mining such a worker's lung-cancer risk. From the papers of Doll et al.,6310 one may estimate the average age of the workers at the middle year of the study to be approximately 58 yr and the average length of time exposed to be approximately 23 yr. However, this does not necessarily imply that their average age at starting such employment was 35 (58 - 23), because "the men regularly change from one type of work to another."l° If the men started working at age 20, their average worktime BaP exposure would be 3,000~23/~58 - 20) ~= 1,816 ng/m3. To express this in constant-exposure terms, we may proceed as follows: Total BaP-carbonization breathed per year = (1,816~9.6~5~('~9) ng = 4.27 ma. That is, 9.6 = m3 of air breathed at work in a working day--8 h at 20 L/min; 5 = working days in a week; and 49 = working weeks in a year. The total air breathed in a year is (15.91~7~49) + (12.48~7~3) = 5,719 m3. That is, 15.91 = [~17.28~5) + (12.48~2~/7 = average m3 breathed per day during a working week, 12.48 = m3 breathed per day during a nonworking day; 17. 98 = m breathed per day during a working day--all these values calculated with assumptions of 20 L/min at work for 8 h, 6 t/min asleep for 8 5, and 10 L/min otherwise. If the gasworkers' exposure is expressed in constant-exposure terms, as though the men breathed such air throughout the day every day, the average BaP-carbonization pollution to which they were exposed is 4.27 mg/5,719 m3 = 747 ng/m3. This led to a 1427 increase in the rate of lung cancer over "background, an estimate roughly 90% of which was caused by the meets smoking habits. If we assume that the relation between duration of exposure and lung cancer risk is the same for gasworks exposure as it is for cigarette- smoking and that the men started work and started to smoke regularly at roughly the same age, we may write (in lung-cancer terms) C-7
10 U.K. cigarettes/d = 0.9 and BaP-carbonization at 747 ng/m3 = 1.42. Those two equations permit us to express BaP-carbonization in terms of U.K. cigarettes as BaP-carbonization at 47.3 ng/m3 = 1 U.K. cigarette. Calculation of the effect of other ages at starting carbonization- work exposure requires more elaborate computation, and the above esti- mate appears to be the best that can be made with the limited data available.30 Note that reasonable changes in the estimate of the proportion of the background lung-cancer rate that was caused by cigarette-smoking have only minor effects on this estimated equivalence. For example, if a figure of 80Z, rather than 90t, is assumed, the equivalence is BaP-carbonization at 42.1 ng/m3 = 1 U.K. cigarette. We assumed in the above calculations that the gasworker breathed 9.6 m3 [(8)(60)(20) L/min] of air at work each working day. The "average" adult breathes roughly half this amount at work. If we assume further that gasworkers and the average 3an breathe similarly at other tines, then the average man breathes 4~543 m of air per year' or 79% (4~543/5,719) as much air as a gasworker. The above equivalent of 47.3 must therefore be divided by this figure to make the exposure applicable to "average" man. Our best estimate is thus finally 59.5, i.e., BaP-carbonization at 59.5 ng/m3 = 1 U.K. cigarette. We estimate from Equation 3 that the lifetime lung-cancer risk associated with exposure to BaP-carbonization at 1 ng/m3 would be 43/100,000. UNITED STATES COKE WORKERS Lloyd and his colleagues21, 32 found in cohort studies of U. S . steelworkers that coke-oven workers experienced a substantial excess risk of lung cancer. These workers, like the British gasworkers, are exposed to the products of coal carbonization. Compared with nonoven workers at the same plants, the coke-oven workers as a group had 2.8 times the lung-cancer mortality rate; and coke-oven workers who had more than 5 yr of "topside" exposure had 6.9 times the lung-cancer mortality rate. No data were given on the smoking habits of these workers or of nonexposed workers, on length of employment, on age, or on average BaP exposure. However, Jackson et al. found average BaP concentrations on the battery roof of a coke-manufacturing plant of 6, 700 ng/m . If this is taken as the BaP exposure of the topside workers, these estimates of lung-cancer risk are remarkably compatible with those from the study of British carbonization workers. The British carbonization workers had a relative risk of lung cancer of 2.42 at a BaP exposure of 3,000 ng/m3, so we may write C-8
Nonexposed British lung-cancer rate = 1.0, Carbonization workers' rate = 2.42, Increment per 1,000 ng/m3 of BaP-carbonizat ion exposure = ~ 2 .42 - 1.0~/3 = 0.47. At the time of these surveys, the age-adjusted U.S. national lung-cancer mortality rate was just half the British rate.l2~37 Taking into account this fac t and the relative risk of 6.9 for the U. S . topside workers, we may wri te Nonexposed U. S. lung-cancer rate = 0. 5, Topside workers' rate = 0.5 x 6.9 = 3.45, Increment per 1, 000 ng/m3 of BaP-carbonization exposure = (3.45 - 0.5~/6.7 = 0.44. The experience of coke-oven workers in the U. S . steel industry is thus in very close agreement with the British data on gasworkers in BaP-exposure terms. LONDON DIESEL-BUS GARAC.F WnRKF.RR the lung-cancer incidence among diesel-bus garage workers employed by the London Jr~s~rt Authority (LTA) has been examined for the period 1950-1974. 1 , ,~ These men were exposed to more diesel emission than other LTA employees, but they showed no greater risk of lung cancer than the other employees. No detailed information on the garage workers' duration of exposure to diesel fumes has been published, but the concentration of smoke was measured inside and outside selected garage s.2, 1 Waller41 concluded that "the indications are that the overall exposure of garage workers to benzota] pyrene during their working 1 Eves would not differ much from those of the general population." The BaP exposure of the U.K. gasworkers discussed above was some 100 times background and was associated with a 14Z; increase in lung-cancer rates. It is therefore hardly surprising that the very small increase over background pollution in a diesel garage (certainly less than a twofold increase) did not produce an epidemiologically measurable effect. Other possible biases in comparing the LTA workers in dif ferent job categories were discussed at length by Harris.15 The study must be considered noninformative, rather than negative; we have discussed it here because it was used as an important data source in recent NRC reportsl5~25 on the impact of particulate emission from diesel-powered light-duty vehicles. OTHER OCCUPATIONALLY EXPOSED GROUPS . . Results of other studies of groups of workers exposed to PAH- containing mixtures were reviewed recently. 25, 34 None of these studies provided evidence of very high exposure; most provided no measure of C-9
actual length or intensity of exposure to PAH-containing mixtures or comparative cigarette-smoking habits. Their results are not useful for purposes of quantitative risk assessment. GENERAL AIR-POLLUTION EXPOSURE Studies of the effects of exposure to general air pollution have been reviewed in numerous reports.439~26334 These reviews have found that lung-cancer rates (as well as rates of cancer at almost all other sites) are higher in urban (i.e., "polluted") than in rural areas (Santodonato et al.,34 Table 6-47~. Interpretation of the increased rates is invariably confounded however, by lack of information on the possible contribution of occupation-induced lung cancer, the possibility of greater accuracy of death certification in urban areas, and, most critically, the lack of detailed information on smoking history. The confounding by occupationally induced lung cancer and more accurate death certification in urban areas is unlikely to be the explanation of most of the urban excess. The confounding by lack of smoking-history information is likely to be the most important.9 We have seen (Figure C-1) that the lung-cancer risk among cigarette-smokers depends strongly on age at starting to smoke, and this holds true even into old age. For valid comparison of lung-cancer rates between urban and rural areas, which allows for smoking-habit differences, it is therefore necessary to know, at a minimum, not only the current smoking habits in the areas being compared, but also the past smoking habits in these areas . In most countries, cigarette-smoking became popular much later in rural than in urban areas; this itself ensures (even allowing for current smoking habits) that lung-cancer rates will be higher in urban than in rural areas of such countries. The above arguments make urban-rural comparisons a very weak basis for evaluating the effect of general air pollution on lung-cancer rates. Moreover, most urban-rural comparisons are of no use for quanti- tative risk-assessment purposes, because they include no estimate of PAH concentrations in the air in the di [ferent areas. LIVERPOOL-NORTH WALES COMPARISON The urban-rural comparison study undertaken by Stocks38 covering the years 1952-1954 in Liverpool (urban) and parts of North Wales (rural) is perhaps unique, in that he not only measured air pollution, but also addressed the issue of long-term smoking habits. The air pollution in the two areas was measured in terms of average BaP concentration over a 2-yr period starting in October 1954: the average BaP concentration in the air was 6.7 ng/m3 in the rural area and 59.2 ng/m3 in the urban area. Stocks addressed the issue of long-term smoking habits by showing that, in men aged 50-59 at the time of the survey in 1953-1955, the urban-rural contrast in smoking habits did not differ from that of 20 yr earlier. C-10