Occupational exposures to polycyclic aromatic hydrocarbons, and respiratory and urinary tract cancers: a quantitative review to 2005

Annals of Oncology 18: 431 446, 2007 doi:10.1093/annonc/mdl172 Published online 25 August 2006 Occupational exposures to polycyclic aromatic hydrocarbons, and respiratory and urinary tract cancers: a quantitative review to 2005 C. Bosetti 1 *, P. Boffetta 2 & C. La Vecchia 1,3 1 Istituto di Ricerche Farmacologiche Mario Negri, Milan, Italy; 2 International Agency for Research on Cancer, Lyon, France; 3 Istituto di Statistica Medica e Biometria, Univesita` degli Studi di Milano, Milan, Italy Received 9 February 2006; revised 21 April 2006; accepted 13 June 2006 Background: Exposure to polycyclic aromatic hydrocarbons (PAHs) has been reported in several industries, including those of the aluminum production, coal gasification, coke production, iron and steel foundries, coal tar and related products, carbon black and carbon electrodes production. Patients and methods: This paper reviews the results from cohort studies conducted on workers exposed to PAHs in these industries, with a focus on cancers of the respiratory and urinary tract. Results: An excess risk from lung/respiratory cancers was found in most industries, the pooled relative risk (RR) being 2.58 (95% CI 2.28 2.92) for coal gasification, 1.58 (95% CI 1.47 1.69) for coke production, 1.40 (95% CI 1.31 1.49) for iron and steel foundries, 1.51 (95% CI 1.28 1.78) for roofers and 1.30 (95% CI 1.06 1.59) for carbon black production. The evidence for cancers of the bladder and of the urinary system is less consistent, with a significant increased risk only for workers in aluminum production (pooled RR = 1.29, 95% CI 1.12 1.49), coal gasification (pooled RR = 2.39, 95% CI 1.36 4.21), and iron and steel foundries (pooled RR = 1.29, 95% CI 1.06 1.57). Conclusions: Increased risks from lung and bladder cancers were found in PAH-related occupations. These were modest in most industries, apart from those for coal gasification, and whether they are due at least partially to some bias or confounding remains open to discussion. Key words: cohort studies, neoplasm, risk, occupational exposure, polycyclic aromatic hydrocarbons, review review introduction Polycyclic aromatic hydrocarbons (PAHs) are a class of chemicals, characterized by the presence of two or more benzene rings, which derive mainly from the incomplete combustion of organic material. These include hundreds of compounds, among which the best known is benzo[a]pirene, often used as a marker of exposure to PAHs [1]. High exposure to PAH mixtures have been reported, mainly in the past, in several industries and occupations, including aluminum production (mainly when the Södeberg process is used), coal gasification, coke production, iron and steel foundries, diesel engine exhaust, coal tar and related products, carbon black and carbon electrodes production and chimney sweeps. The epidemiological evidence from workers occupationally exposed to PAHs in various industries and occupations has been reviewed by the International Agency for Research on Cancer (IARC), which classified some of *Correspondence to: Dr C. Bosetti, Istituto di Ricerche Farmacologiche Mario Negri, Via Eritrea 62 20157 Milan, Italy. Tel: +39-0239014526; Fax: +39-0233200231; E-mail: bosetti@marionegri.it these industries as carcinogenic to humans [2 7], mainly for an increased risk of lung, laryngeal, skin, kidney and urinary bladder cancer. This evidence has been also included in a comprehensive review by Boffetta et al. [8]. This paper reviews the main results from cohort studies conducted on workers from PAH-related occupations, with emphasis on study results reported after the review by Boffetta et al. [8]. Particular attention is given to cancers of the respiratory and urinary tract, since these are the sites most consistently associated with PAH exposures. Skin cancer has not been considered, since ascertainment of incidence or mortality from this neoplasm is problematic. Exposure to diesel engine exhaust in the transport industry and related occupations, which has been related to some excess risk of lung and bladder cancer [8 10], is also not included in the present review, in consideration of the heterogeneous and highly variable working conditions of these groups of workers and the consequent difficulty of obtaining reliable overall risk estimate. We reviewed, in detail, cohort studies published between 1997 and 2005, while for cohort studies included in the review by Boffetta et al. [8] we summarized only the main ª 2006 European Society for Medical Oncology

Annals of Oncology Table 1. Cohort studies of aluminum production workers Authors and publication year Country and study population employment follow-up, outcome Cancer Observed deaths/cases SMR/SIR (95% CI) Milham [12] USA 1946 62 1946 76 Respiratory tract 35 1.2 (0.8 1.6) 2103 male workers Mortality Bladder 1 0.4 (0.1 2.1) at one prebake plant Kidney 2 0.9 (0.1 3.2) Giovanazzi and Italy 1965 79 Lung 4 0.8 (0.2 2.0) D Andrea [13] 494 workers at one Söderberg plant Mortality Larynx 1 2.5 (0.3 14) Andersen et al. a [14] Ronneberg et al. a,b [15] Ronneberg et al. a,b [16] Norway 1953 70 1953 79 Lung 57 1.6 (1.2 2.1) 7410 males at four Mortality, Larynx 7 1.2 (0.5 2.5) plants incidence Bladder 26 1.2 (0.8 1.7) Kidney 18 1.2 (0.7 1.9) Norway 1922 75 1953 91 Lung 1137 male workers Mortality, <3 years employment 20 2.7 (1.6 4.1) at one Prebake plant incidence 3 years employment 19 1.2 (0.7 1.8) Bladder 5 <3 years employment 14 1.3 (0.4 3.0) 3 years employment 1.6 (0.9 2.7) Norway 2647 short term male workers + 2888 production male workers + 373 maintenance male workers 1946 1953 93 Incidence Short-term workers Lung Upper respiratory tract 41 6 1.5 (1.1 2.1) 0.9 (0.3 1.9) Bladder 8 0.6 (0.2 1.1) Kidney 9 1.1 (0.5 2.2) Production Lung 42 1.0 (0.7 1.3) Upper respiratory tract 12 1.1 (0.6 2.0) Bladder 23 1.0 (0.6 1.5) Kidney 13 1.1 (0.6 1.8) Maintenance Lung 10 2.1 (1.0 2.9) Upper respiratory tract 2 1.7 (0.2 6.0) Bladder 5 2.0 (0.7 4.7) Kidney 1 0.7 (0.02 4.1) Romundstad et al. a Norway 1919 95 1953 95 Lung 27 0.9 (0.6 1.3) [17] 1790 male workers Incidence Bladder 23 1.3 (0.8 1.9) at one Söderberg plant Kidney 10 1.1 (0.5 2.0) (since 1939) Romundstad et al. a Norway 1954 95 1962 95 Lung [18] 5627 male workers Incidence <3 years employment 15 1.2 (0.7 2.0) at two Söderberg plants 3 years employment 46 0.9 (0.7 1.2) Bladder <3 years employment 5 0.8 (0.3 1.8) 3 years employment 36 1.4 (1.0 1.9) Kidney <3 years employment 3 0.8 (0.2 2.4) 3 years employment 12 0.9 (0.4 1.5) Romundstad et al. Norway 1953 96 Lung 189 1.0 (0.9 1.2) [19] 11 103 male workers Incidence Larynx 24 1.3 (0.8 1.9) at six plants Bladder 130 1.3 (1.1 1.5) Kidney 55 1.1 (0.8 1.4) Rockette and Arena USA 1946 77 1946 77 Lung (Prebake) 161 1.0 (0.8 1.1) [20] 21 829 male workers Mortality Lung (Söderberg) 64 0.9 (0.7 1.1) at 14 plants Bladder (Prebake) 11 0.7 (0.4 1.3) Bladder (Söderberg) 8 1.6 (0.7 3.2) Kidney (Prebake) 19 1.5 (0.9 2.3) Kidney (Söderberg) 3 0.5 (0.1 1.6) Gibbs [21] Canada 1950 51 1950 77 Lung 131 1.3 (1.1 1.6) 5406 male workers at two Söderberg plants Mortality Bladder 13 1.2 (0.6 2.0) 432 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology review Table 1. (Continued) Authors and publication year Country and study population employment follow-up, outcome Cancer Observed deaths/cases SMR/SIR (95% CI) Mur et al. [22] France 1950 76 1950 76 Lung 37 1.1 (0.9 1.5) 6455 male workers Mortality Bladder 7 2.1 (1.0 3.7) at 11 plants Moulin et al. c [23] France 1950 94 1968 94 Lung 19 0.6 (0.4 1.0) 2133 male workers Mortality Larynx 7 1.1 (0.5 2.3) at one plant Bladder 7 1.8 (0.7 3.6) Spinelli et al. [24] Canada 1954 85 1954 85 Lung 37 1.0 (0.7 1.3) 4213 male workers Mortality, Larynx 6 1.7 (1.1 2.6) at one Söderberg plant incidence Bladder 16 1.5 (0.7 3.0) Kidney 7 1.2 (0.6 2.3) Carta et al. [25] Italy 1971 80 1971 90 Lung 3 0.6 (0.2 2.0) 1148 workers at one plant Mortality Carta et al. d [26] Italy 1972 80 1972 01 Lung 11 0.7 (0.4 1.3) 1152 male workers at one Prebake plant Mortality Bladder 3 0.8 (0.3 2.4) SMR, standardized mortality ratio; SIR, standardized incidence ratio; CI, confidence interval. a Data included in Romundstad et al. [19]. b Update of data from one plant included in Andersen et al. [14]. c Update of data from one plant included in Mur et al. [22]. d Update of Carta et al. [25]. Table 2. Cohort studies of coal gasification workers Authors and publication year Country and study population employment follow-up, outcome Cancer Observed deaths/cases SMR/SIR (95% CI) Doll et al. [27] UK 1953 65 1953 65 Lung 99 1.8 (1.5 2.2) 2449 male workers Mortality Bladder 10 2.3 (1.1 4.2) in four companies Manz [28] Germany 1953 77 Respiratory tract 63 3.7 5405 male workers Mortality Urinary tract 6 6.1 in one company Hansen et al. [29] Denmark Lung 7 8.9 47 male workers, 141 Mortality non exposed controls Gustavsson and Sweden 1965 72 1966 86 Lung 4 0.8 (0.2 2.1) Reuterwall [30] 295 male workers Mortality, Bladder 2 2.9 (0.3 10) in one company incidence Berger and Germany 1900 89 1953 89 Lung 78 2.9 (2.3 3.6) Manz [31] 789 male workers in the furnace of one company Mortality SMR, standardized mortality ratio; SIR, standardized incidence ratio; CI, confidence interval. results. We also calculated quantitative estimates of the risk of selected cancers based on meta-analyses of results from all available studies. patients and methods The studies included in this quantitative review were original cohort investigations published up to December 2005 on workers from selected industries characterized by high exposure to PAHs, and providing information on the risk of lung, bladder and kidney cancers, or more in general of neoplasms of the respiratory and urinary tract. They were identified through searches of the MEDLINE database, using the keywords polycyclic aromatic hydrocarbons, neoplasm, risk, occupational exposure or cohort studies. Papers were also searched among those quoted in the retrieved studies. A total of 62 papers were identified, including the results of 57 cohort studies, whose main characteristics and results are described in Tables 1 7. To provide a quantitative overall estimate of the standardized mortality ratios (SMR) or incidence ratios (SIR) for selected cancers in relation to various industries and occupations, the number of cancer deaths/cases Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 433

Annals of Oncology Table 3. Cohort studies of lung cancer in coke production workers Authors and publication year Country and study population employment follow-up, outcome Job Observed deaths/cases SMR/SIR (95% CI) Reid and Buck [32] UK 1949 54 1950 54 Coke oven workers 14 1.4 (0.8 2.3) 8000 male workers at one plant Mortality Sakabe et al. [33] Japan 1949 73 1949 73 Coke oven workers 15 1.3 (0.7 2.1) 2178 male retired workers Mortality from 11 companies Davies [34] UK 1954 65 1954 65 Coke oven workers 8 0.8 (0.4 1.6) 610 male workers at two steel plants Mortality Hurley et al. [35] UK 1966 67 1966 80 All workers 167 1.2 (1.0 1.4) 2842 male workers at 14 steel plants Mortality Steel plants 63 1.2 (0.9 1.5) 3952 male workers at 13 coking plants Coking department 104 1.2 (1.0 1.4) Wu [36] China 1930 65 1971 82 All workers 93 2.6 (2.1 3.1) 21 995 workers at 19 coking plants Mortality Coking department 53 4.4 (3.3 5.8) Preparing department 12 2.5 (1.3 4.3) Swaen et al. [37] The Netherlands 1954 84 Coke oven workers 62 1.3 (1.0 1.7) 5639 male workers at three plants 1945 69 Mortality By-product workers 104 1.0 (0.8 1.2) Franco et al. [38] Italy 1960 85 1960 90 Coke oven workers 19 1.7 (1.0 2.7) 538 workers in one plant Mortality Chau et al. [39] France 1963 82 1963 87 Coke oven workers 2 1.8 (0.2 6.3) 536 male retired workers Mortality Near oven workers 8 2.5 (1.1 5.0) from two plants Repair maintenance 6 4.3 (1.6 9.4) workers By-product workers 2 2.4 (0.3 8.6) Costantino et al. [40] USA and Canada 1951 55 1951 82 Coke oven workers 255 2.0 (1.6 2.3) 15 818 male workers in 12 companies Mortality Bye et al. [41] Norway 1964 88 1962 93 All workers 7 0.8 (0.3 1.7) 888 male workers Incidence SMR, standardized mortality ratio; SIR, standardized incidence ratio; CI, confidence interval. observed, and the SMR/SIR were abstracted from each paper. The expected number of deaths/cases in each study was also abstracted or obtained from the ratio of observed deaths/cases over SMR/SIR, and the overall SMR for each neoplasm of interest was calculated as an unweighted ratio of the sum of observed and expected deaths/cases (Table 8). Pooled relative risks (RR) and the corresponding 95% confidence intervals (CI) were also computed as a weighted average of the SMR/SIR, using the inverse of the variance of the logarithm of the SMR/SIR as weight (fixed-effects model [10], Table 8 and Figures 1 7). Chi-square tests for heterogeneity were used to evaluate the consistency of findings between studies [11]. For some major industries, the results of the meta-analysis were presented graphically, plotting the SMR/SIR as a black square, whose size was inversely proportional to the variance of the logarithm of the SMR/SIR. For various studies, the 95% CI differed slightly from those published in the original papers because of rounding. For studies which presented results for different categories of workers within the same industry, a single pooled estimate was plotted. Similarly, a single pooled estimate combining studies with less than 10 cancer cases was plotted. Diamonds were used to plot the pooled RR for all studies; their centre represents the RR and their extremes the 95% CI. results aluminum production Table 1 gives information from 15 papers on nine cohorts of aluminum production workers [12 26]. Over the period 1997 2005, five papers were published from Norway [16 18], France [23] and Italy [26]. The data from the three Norwegian studies [16 18] were then included in a subsequent re-analysis [19]. This re-analysis included the results of the cohorts of six aluminum plants from Norway, two of which started to operate before 1920, and the remaining from the period between 1947 and 1958. In total, that study included 11 103 men employed for more than 3 years, and followed-up for cancer incidence between 1953 and 1996. With reference to lung cancer, no excess risk was observed (189 observed cases versus 183.1 expected, corresponding to a SIR of 1.0; 95% CI 0.9 1.2), nor a trend with the cumulative dose of exposure to PAHs. For cancer of the bladder, 130 cases were observed versus 101.7 expected, corresponding to a SIR of 1.3 (95% CI 1.1 1.5). There was also a significant trend in bladder cancer risk in relation to the cumulative dose of PAHs, with a RR of 2 for exposures 2000 lg/m 3 for 30 or more years. Thus, this large reanalysis of all the Norwegian studies of aluminum production, including more than 1500 cancer cases and with a long period of follow-up, confirmed an excess risk for cancer of the bladder, but did not show any association for lung cancer. A mortality study conducted by the French National Institute for the Research and Safety [23] updated the data on workers from one of the aluminum reduction plants considered in the paper by Mur et al. [17]. That study 434 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology review Table 4. Cohort studies of iron and steel foundry workers Authors and publication year Country and study population employment follow-up, outcome Cancer Observed deaths/cases SMR/SIR (95% CI) Koskela et al. [42] Finland 1950 72 1950 72 Lung 10 2.1 (1.0 3.8) 1233 male workers Mortality (employed for five or more years) Gibson et al. [43] Canada 1967 1967 77 Lung 21 2.5 (1.5 3.8) 439 workers in one company Mortality Breslin [44] USA 1953 1953 70 Lung 34 1.0 (0.7 1.4) 2167 male workers at seven steel foundries Mortality Bladder 3 1.0 (0.2 2.8) Kidney 4 1.6 (0.4 4.1) Decouflé [45] USA 1938 67 1938 67 Respiratory tract 29 1.3 (0.8 1.8) 2861 male workers at one iron foundry Mortality Bladder 3 1.1 (0.2 3.1) Kidney 3 1.6 (0.3 4.6) Andjelkovich et al. USA 1950 79 1950 84 Lung 139 1.3 (1.1 1.5) [46 48] 8147 male workers at one iron foundry Mortality Larynx 4 0.7 (0.2 1.8) Bladder 8 1.0 (0.4 2.0) Kidney 9 1.1 (0.5 2.1) Hansen [49] Denmark 1970 1970 80 Lung 9 1.4 (0.6 2.6) 632 molders Mortality Bladder 6 9.0 (3.3 19) Sherson et al. [50] Denmark 1967 74 1967 85 Lung 85 1.2 (1.0 1.5) 3377 male workers at one foundry Incidence Bladder 32 1.1 (0.8 1.6) Rotimi et al. [51] USA 1952 86 1973 86 Lung 72 1.2 (0.9 1.5) 5540 male workers at one iron foundry Mortality Sorahan et al. [52] UK 1946 65 1946 90 Lung 551 1.5 (1.3 1.6) 10 438 male workers at nine steel foundries Mortality Larynx 12 1.3 (0.7 2.3) Bladder 37 1.1 (0.8 1.5) Kidney 24 1.3 (0.9 2.0) Moulin et al. [53] France 1968 92 1946 90 Lung 54 1.2 (0.9 1.6) 4288 male and 609 female workers Mortality Larynx 17 1.5 (0.9 2.4) at one stainless and alloyed steel plant Bladder 10 1.7 (0.8 3.2) SMR, standardized mortality ratio; SIR, standardized incidence ratio; CI, confidence interval. included 2133 men employed for at least 1 year between 1950 and 1994, and followed up between 1968 and 1994. Overall, mortality from cancer of the lung was below unity (SMR = 0.6, 95% CI 0.4 1.0, on the basis of 19 observed deaths). There was no excess lung cancer in any of the specific jobs where exposure to PAHs was more likely, nor trends with duration of exposure and time since first exposure. With reference to bladder cancer, the SMR was 1.8, on the basis of seven observed deaths (95% CI 0.7 3.6). Six deaths occurred in jobs with more probable exposure to PAH (SMR = 2.15, 95% CI 0.79 4.68). However, no increase in risk was observed for longer duration of exposure and time since first exposure. The Italian study [26] was a 10-year mortality update of a cohort of workers from a Prebake aluminum smelter [18], which included 1152 men employed for at least 1 year in the foundry between 1972 and 1980, and followed-up until the end of 2001. No excess mortality was observed for cancer of the lung (11 observed deaths versus 15.8 expected, SMR = 0.7, 95% CI 0.4 1.3), nor of the bladder (three observed deaths versus 3.8 expected, SMR = 0.8, 95% CI 0.3 2.4). Furthermore, no difference in risk was observed when workers were classified in three groups with increasing level of exposure to PAHs. In eight cohorts of aluminum production workers, 653 lung cancer cases were observed versus 644.9 expected, with a pooled RR of 1.03 (95% CI 0.95 1.11, Table 8). Considering all cancers of the respiratory system, mainly lung cancer, a total of 688 cases were observed compared with 674.1 expected, with a pooled RR of 1.03 (95% CI 0.96 1.11, Table 8 and Figure 1). For bladder cancer, the eight available studies included a total of 196 cases observed versus 155.7 expected, with a pooled RR of 1.29 (95% CI 1.12 1.49, Table 8 and Figure 2). With reference to kidney cancer, the cases observed in four studies were 86 versus 76.7 expected, with a pooled RR of 1.15 (Table 8). coal gasification In the past, coal gasification implied high exposures to PAHs, as to other carcinogenic and toxic substances. Five cohort studies are available on exposures occurred in the past (before the 1970s) whose main results are summarized in Table 2 [27 31]. No study on workers employed in Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 435

Annals of Oncology Table 5. Cohort studies of workers exposed to bitumen fumes, coal tar and related products Authors and publication year Country and study population employment follow-up, outcome Cancer Observed deaths/cases SMR/SIR (95% CI) Tar distillation Maclaren and Hurley [54] UK 1967 1967 83 Lung 12 1.6 (0.8 2.8) 255 male workers at 4 plants Mortality Bladder 3 4.3 (0.9 12.5) Moulin et al. [55] France 1970 84 1970 84 Lung 5 0.7 (0.2 1.6) 963 workers at 1 plant Mortality Larynx 0 0.0 (0.0 1.5) Bladder 0 0.0 (0.0 4.6) Swaen et al. [56] The Netherlands 1947 80 1947 88 Lung 48 1.2 (0.9 1.6) 907 workers at 1 plant Mortality Larynx 0 Bladder 2 0.5 (0.1 2.0) Kidney 0 Shale oil extraction Bogovski [57] Estonia Lung <1.0 P>0.05 2003 workers Incidence Miller et al. [58] UK 1950 62 1950 82 Lung 84 0.9 (0.7 1.1) 6064 male workers Mortality Bladder 8 0.8 (0.4 1.7) Kidney 4 1.0 (0.4 2.7) Creosote exposure Karlehagen et al. [59] Norway and Sweden 1950 75 1953 87 Lung 13 0.8 (0.4 1.3) 922 male wood impregnators at 13 plants Incidence Bladder 10 1.1 (0.5 2.0) Wong and Harris [60] USA 1979 99 1979 01 Lung 34 1.3 (0.9 1.9) 2179 workers in 11 wood-treating plants Mortality Larynx 2 1.6 (0.2 6.6) Bladder 0 0.0 (0.0 3.0) Kidney 3 1.9 (0.4 5.5) Roofers Hammond et al. [61] USA 1960 1960 71 Lung 99 1.6 (1.3 1.9) 5939 male roofer union members Mortality Bladder 13 1.7 (0.9 2.9) Swaen et al. [56] The Netherlands 1947 80 1947 88 Larynx 1 1.4 (0.02 7.9) 866 roofers Mortality Lung 39 1.3 (0.9 1.8) Bladder 3 1.1 (0.2 3.4) Kidney 2 1.0 (0.1 3.8) Asphalt workers Hansen [62] Denmark 1959 80 1959 84 Lung 27 3.4 (2.3 5.0) 679 asphalt mastic workers Incidence Larynx 3 4.4 (0.9 13) Bladder 5 1.6 (0.5 3.6) Boffetta et al. [63] Six European countries + Israel 1913 1999 1953 2000 Lung 795 1.1 (1.0 1.2) 79 822 asphalt workers Mortality Larynx 42 1.0 (0.7 1.3) Bladder 104 1.0 (0.8 1.2) Kidney 80 0.8 (0.7 1.0) SMR, standardized mortality ratio; SMR, standardized incidence ratio; CI, confidence interval. this manufacture was published after the review by Boffetta et al. [8]. Overall, 188 deaths from lung cancer were reported in the five cohorts of coal gasification workers, compared with 87.7 expected, with a pooled RR of 2.29 (95% CI 1.98 2.64, Table 8). Considering all cancers of the respiratory system, 251 deaths were observed versus 104.7 expected (pooled RR = 2.58, 95% CI 2.28 2.92, Table 8 and Figure 3). All the estimates were significantly above unity, although a certain degree of heterogeneity was found. The number of deaths observed from cancer of the bladder in the two cohorts were 12 versus 5.0 expected, corresponding to a pooled RR of 2.39 (95% CI 1.36 4.21); 18 deaths from all cancers of the urinary system were recorded, versus 6.0 expected (pooled RR = 3.27, 95% CI 2.06 5.19, Table 8). coke production With reference to coke production, there are 10 cohorts of workers [32 41] (Table 3), among which only a small Norwegian investigation [41] was published after the review by Boffetta et al. [8]. That study was based on a cohort of 888 workers of one coking plant that operated between 1964 and 1988, followed-up for incidence of cancer and other diseases until 1993. No excess was observed for cancer of the lung (seven observed cases versus 8.5 expected, SIR = 0.82) and of the bladder (two observed cases versus 2.4 expected, SIR = 0.82). However, an excess risk was observed in relation to the cumulative dose of PAHs (SIR = 3.60 for 150 lg/m 3 year of PAH in the overall cohort, and 4.82 for 15 years of 436 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology review Table 6. Cohort studies of workers of carbon black production Authors and publication year Country and study population employment follow-up, outcome Cancer Observed deaths/cases SMR (95% CI) Hodgson and Jones [66] UK 1947 80 1947 80 Lung 25 1.5 (1.0 2.2) 1422 male workers at 5 plants Mortality Bladder 3 2.5 (0.5 7.3) Sorahan et al. a [67] UK 1956 74 1951 96 Lung 61 1.7 (1.3 2.2) 1147 male manual workers for Mortality Larynx 2 2.2 (0.3 7.9) 12 or more months at 5 plants Bladder 6 1.7 (0.6 3.8) Robertson and Inman [65] USA 1935 74 1935 94 Respiratory tract 34 0.8 (0.6 1.1) 5 plants Mortality SMR, standardized mortality ratio; CI, confidence interval. a Update of Hodgson and Jones [66]. Table 7. Cohort studies of workers of carbon electrode manufacture Authors and publication year Teta et al. [68] Country and study population USA 2219 male workers at 11 plants from Union Carbide Corporation employment follow-up, outcome 1974 1974 1983 Mortality Cancer Respiratory tract Urinary tract Observed deaths/cases 29 4 SMR/SIR (95% CI) 0.9 (0.6 1.2) 0.9 (0.3 2.4) Moulin et al. [69] France 1975 1975 1985 Lung 7 0.8 (0.3 1.6) 1302 workers Incidence (cohort A) Bladder 0 0 (0.0 1.9) Urinary system 2 1.7 (0.2 6.2) Moulin et al. [69] France 1957 1957 1984 Lung 13 1.2 (0.6 2.0) 1115 workers Mortality (cohort B) Bladder 3 1.9 (0.4 5.7) Gustavsson et al. [70] Sweden 1968 1988 1969 1989 Lung 2 1.7 (0.2 6.1) 901 workers (807 males, Mortality Bladder 1 8.3 (0.2 46.4) 94 females) employed for >3 months Donato et al. [71] Italy 1945 1971 1955 1996 Lung 34 0.8 (0.5 1.1) 1006 male workers Mortality Larynx 4 0.8 (0.2 2.0) employed for >1 year Bladder 7 1.0 (0.4 2.1) Mori [72] Japan 1951 74 1951 88 Lung 9 2.6 (1.2 5.0) 332 male workers employed for more than 5 years Merlo et al. [73] Italy 1950 1989 1950 1997 Lung 32 1.0 (0.7 1.4) 1291 male workers employed for >1 year Mortality Bladder 5 1.1 (0.4 2.5) SMR, standardized mortality ratio; SIR, standardized incidence ratio; IC, confidence interval. latency). The result of this small study, with a short period of follow-up, should however be interpreted with caution. Overall, in 10 cohort studies 762 cases of lung cancer were observed, compared with 512.1 expected, with a pooled RR of 1.58 (95% CI 1.47 1.69, Table 8 and Figure 4). A considerable heterogeneity of risk estimates across various studies was, however, observed, particularly between the four larger studies [20 23], which gave RRs of 2.0, 1.2, 1.1 and 2.6, respectively. Some studies reported some excess risk for other neoplasms, such as laryngeal, liver, prostate or kidney cancers, but overall data do not provide a consistent evidence of any excess of risk. Likewise, for cancer of the bladder, no evidence of an increased risk was observed in any of the studies. iron and steel foundries Iron and steel foundry workers are another important group of workers exposed to considerable levels of PAHs, mainly in the past. The main results of 10 cohort studies on these workers are shown in Table 4 [42 53]. After the review by Boffetta et al. [8], a mortality study of a cohort of 4288 men and 609 women from a French stainless and alloyed steel plant was published [51]. Workers were employed between the beginning of 1968 and the end of 1991, and were followed-up to the end of 1992. No excess mortality from lung cancer was reported in the overall cohort (SMR = 1.2, 95% CI 0.9 1.6, on the basis of 54 observed deaths). The SMR of Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 437

Annals of Oncology Table 8. Overall standardized mortality ratio (SMR) and pooled relative risk (RR) with 95% confidence interval (CI) for exposure to polycyclic aromatic hydrocarbons in various industries and occupations Industry, cancer site No. of cohorts Observed/Expected SMR Pooled RR a Test for heterog. P-value Aluminum production Lung 8 653/644.9 1.01 1.03 (0.95 1.11) 0.04 Respiratory tract c 9 688/614.1 1.02 1.03 (0.96 1.11) 0.048 Bladder 8 196/155.7 1.26 1.29 (1.12 1.49) 0.31 Kidney 4 86/76.7 1.12 1.15 (0.93 1.42) 0.45 Coal gasification Lung 4 188/87.7 2.14 2.29 (1.98 2.64) <0.0001 Respiratory tract c 5 251/104.7 2.40 2.58 (2.28 2.92) <0.0001 Bladder 2 12/5.0 2.38 2.39 (1.36 4.21) 0.77 Urinary tract c 3 18/6.02 2.99 3.27 (2.06 5.19) 0.17 Coke production Lung 10 762/512.1 1.49 1.58 (1.47 1.69) <0.0001 Iron and steel foundry Lung 9 975/703.7 1.39 1.40 (1.32 1.49) 0.007 Respiratory tract c 10 1004/726.0 1.38 1.40 (1.31 1.49) 0.012 Bladder 7 99/83.0 1.19 1.29 (1.06 1.57) <0.001 Kidney 4 40/31.0 1.29 1.30 (0.95 1.77) 0.91 Tar distillation Lung 3 65/54.6 1.19 1.21 (0.95 1.55) 0.30 Bladder d 3 5/6.1 0.82 Creosote Lung 2 47/42.4 1.11 1.14 (0.85 1.51) 0.14 Roofers Lung 2 138/91.9 1.50 1.51 (1.28 1.78) 0.27 Bladder 2 16/10.4 1.54 1.57 (0.96 2.56) 0.50 Asphalt workers Lung 2 822/730.7 1.12 1.14 (1.07 1.22) <0.0001 Bladder 2 109/107.1 1.12 1.02 (0.85 1.23) 0.31 Carbon black production Lung 2 95/78.4 1.21 1.30 (1.06 1.59) <0.0001 Carbon electrode manufacture Lung 6 97/101.6 0.96 1.00 (0.82 1.23) 0.037 Respiratory tract c 7 126/135.7 0.93 0.96 (0.81 1.15) 0.053 Bladder 4 16/13.0 1.23 1.35 (0.83 2.20) 0.22 Urinary tract e 6 20/17.3 1.15 1.29 (0.85 1.95) 0.41 a Calculated as a weighted average of the SMRs, using the inverse of the variance as weight. b Some overlapping of data from Mur et al. [22] and Moulin et al. [23] is possible. c Including lung, and other respiratory cancers not specified. d The pooled RR was not estimated, given the low number of events. e Including bladder and other urinary cancers not specified. bladder cancer was 1.7 (95% CI 0.8 3.2) on the basis of 10 observed deaths. In a case control analysis of lung cancer nested in the cohort, a RR of 1.9 (95% CI 1.0 3.7) was reported for exposure to PAHs, with a significant trend with duration and cumulative dose of exposure [51]. Ten cohort studies available on workers in iron and steel foundries included a total of 975 lung cancer cases observed compared with 703.7 expected, with a pooled RR of 1.40 (95% CI 1.32 1.49, Table 8). Considering all respiratory neoplasms, there were 1004 observed cases versus 726.0 expected, with a pooled RR of 1.40 (95% CI 1.31 1.49, Table 8 and Figure 5). Cancers of the bladder were 99 versus 83.0 expected based on seven studies, with a pooled RR of 1.29 (95% CI 1.06 1.57, Table 8 and Figure 6). A significant heterogeneity between studies was observed, mainly due to a small cohort, which reported a SMR of 9 [25]. For kidney cancer, 40 cases were observed versus 31.0 expected (pooled RR = 1.30, 95% CI 0.95 1.77, Table 8). bitumen fumes, coal tar and related products In relation to coal tar and related products, there were 11 cohorts of workers, three of tar distillation [54 56], two of shale oil extraction [57, 58], two of creosote [59, 60], two of roofers [56, 61] and two of asphalt workers [62, 63] (Table 5). 438 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology review Figure 1. Relative risks (RR) of lung cancer (including other respiratory cancers not specified) among aluminum production workers, and corresponding confidence intervals (CI), by study and overall. A mortality study of a cohort of 907 tar distillers and 899 roofers employed in a Dutch industry [56] was published after the review by Boffetta et al. [8]. Workers were employed for at least 6 months between 1947 and 1980, and were followed-up until the beginning of 1988. A non-significant excess mortality from lung cancer was reported for tar distillers, with a total of 48 observed deaths versus 40.6 expected, corresponding to a SMR of 1.2 (95% CI 0.9 1.6). For cancer of the bladder, the SMR was of 0.5 (95% CI 0.06 1.87), on the basis of two observed deaths versus 3.6 expected. A retrospective cohort study of workers of creosote-based wood treatments [60] was published after the review by Boffetta et al. [8]. This included over 2000 individuals who were employed at 11 wood-treating plants in the USA between 1979 and 1999, followed-up for mortality up to 2001. A modest excess risk for lung cancer mortality was observed (SMR = 1.3, 95% CI 0.9 1.8, based on 38 deaths observed). The SMR for laryngeal cancer was 1.6 (95% CI 0.2 6.6), based on two deaths observed. No deaths from cancer of the bladder and other urinary organs were found, and three deaths were observed from kidney cancer (SMR = 1.9, 95% CI 0.4 5.5). Similar findings were found among hourly employees. Analyses by length of employment and latency among hourly employees showed no clear pattern for both respiratory tract and kidney cancers. With reference to roofers, in the study by Swaen et al. [56] there were 39 deaths from lung cancer compared with 29.7 expected, corresponding to a non-significant SMR of 1.3 (95% CI 0.9 1.8). The SMR for bladder cancer was close to unity (SMR = 1.1, 95% CI 0.2 3.4), with three observed deaths versus 2.6 expected. The results of a large cohort study on asphalt workers were recently published [63]. The study was conducted in seven European countries and Israel, and was coordinated by the IARC. It included a total of 79 822 workers, of whom 29 820 were asphalt workers exposed to bitumen fumes, 32 245 constructors not exposed to bitumen fumes, and 17 757 other workers not classifiable as bitumen workers. The workers had been employed for the first time in a variable period between 1913 and 1999. The mean duration of follow-up (which started at variable times in various countries) was 16.7 years, ranging from 11.7 years in France and Germany to 21.9 years in Norway. Overall, in the asphalt workers there were 795 lung cancer deaths observed versus 742.4 expected, corresponding to a SMR of 1.1 (95% CI 1.0 1.2). The SMR for cancer of the lung was higher in the asphalt workers exposed to bitumen (SMR = 1.2, on the basis of 330 observed deaths), than in building or ground constructors (SMR = 1.0, on the basis of 249 observed deaths) and in other workers (SMR = 1.0, on the basis of 216 observed deaths). No excess risk was observed for laryngeal cancer (SMR = 1.0, on the Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 439

Annals of Oncology Figure 2. Relative risks (RR) of bladder cancer among aluminum production workers, and corresponding confidence intervals (CI), by study and overall. Figure 3. Relative risks (RR) of lung cancer (including other respiratory cancers not specified) among coal gasification workers, and corresponding confidence intervals (CI), by study and overall. 440 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology review Figure 4. Relative risks (RR) of lung cancer among coke production workers, and corresponding confidence intervals (CI), by study and overall. basis of 42 deaths). For bladder cancer, there were 104 deaths observed versus 104.0 expected (SMR = 1.0, 95% CI 0.8 1.2). Similarly, no excess mortality from kidney cancer was found in asphalt workers (SMR = 0.8 95% CI 0.7 1.0, on the basis of 80 deaths observed). The results of this large study suggest, therefore, that asphalt workers had no excess of mortality from cancer of the respiratory and urinary tract. A subsequent analysis of the same cohort of asphalt workers, based on a semi-quantitative job-exposure matrix [64], reported no excess mortality from lung cancer in workers exposed to bitumen fumes (SMR = 1.1), as well as in those exposed to coal tar (SMR = 1.1). Conversely, an excess mortality from lung cancer was found in workers exposed to diesel exhaust (SMR = 1.31). Overall, the pooled RR for lung cancer was 1.21 (95% CI 0.95 1.55) in workers of tar distillation, 1.14 (95% CI 0.85 1.51) in those exposed to creosote, 1.51 (95% CI 1.28 1.78) in roofers and 1.14 (95% CI 1.07 1.22) in asphalt workers (Table 8). With reference to bladder cancer, the pooled RR was 1.82 (95% CI 0.76 4.37) in workers of tar distillation, 1.57 (95% CI 0.96 2.56) in roofers and 1.02 (95% CI 0.85 1.23) in asphalt workers (Table 8). carbon black manufacture Three papers are available on two cohorts of carbon black production workers [65 67] (Table 6). A US study did not show any significant excess mortality from cancers of the respiratory system [65]. Conversely, an excess mortality from lung and bladder cancer was reported in an English study [66], and was confirmed in a recent update of the same cohort [67]. The excess of mortality from lung cancer seemed limited, however, to two of the five plants considered [67]. Moreover, no relation was found with time since first employment, and internal analyses did not evidence any association with cumulative dose of exposure to carbon black. Altogether, the data from two cohort studies of carbon black production reported 95 deaths from lung cancer versus 78.4 expected (SMR = 1.30, 95% CI 1.06 1.59, Table 8). However, the results from these two cohorts were not consistent. carbon and graphite electrode manufacture The main results from seven cohort studies that reported mortality in carbon electrode workers are given in Table 7 [68 73]. After the revision by Boffetta et al. [8], a historical cohort study was conducted on 1006 male workers employed for at least 1 year between 1945 and 1971 in an Italian plant, and followed-up for mortality between 1955 and 1996 [71]. With reference to specific cancer sites, 34 deaths from lung cancer were reported compared with 44.2 expected (SMR = 0.8, 95% CI 0.5 1.1), and seven deaths from bladder cancer compared to 6.7 expected (SMR = 1.0, 95% CI 0.4 2.1). Analyses of mortality by time since first employment showed a higher risk, although not significant, of lung cancer in subjects with Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 441

Annals of Oncology Figure 5. Relative risks (RR) of lung cancer (including other respiratory cancers not specified) among iron and steel foundry workers, and corresponding confidence intervals (CI), by study and overall. 20 years since first employment. The risk, however, was inversely related to the duration of employment. A small cohort from Japan [72] was based on 322 male workers employed for more than 5 years in a man-made graphite electrode manufacturing plant in the period 1951 1974, and followed-up for mortality up to 1988. In that cohort, nine deaths from lung cancer were observed compared with 3.4 expected, corresponding to a SMR of 2.6 (95% CI 1.2 5.0). The excess risk was found in separate sub-period analysis. No deaths from bladder cancer were observed, and therefore no excess risk for any other neoplasm. Another cohort study was conducted on 1291 males employed between 1950 and 1989 in a graphite electrode production plant of Val Camonica, northern Italy, and followed-up until the end of 1997 [73]. The Italian population was used as reference for the period 1950 1989 and the Lombardy population was used for the period 1990 1997. That study did not report any excess of mortality from cancer of the lung (SMR = 1.0) or of the bladder (SMR = 1.1). Analyses stratified by age at death, time since first employment and duration of employment did not show any consistent pattern either for lung or for bladder cancer. The data of a retrospective cohort from China are also available [74]. That cohort included 6635 male workers employed for more than 15 years during the period 1970 1985 in seven factories, including the carbon plants and the pot room and carbon department in an aluminum reduction plant. An excess mortality from lung cancer was observed (50 deaths observed versus 23.1 expected, SMR = 2.16) with stronger excess in the highly exposed group (SMR = 4.3, based on 26 deaths observed). This data, however, included workers of carbon product manufacturing, along with those of other activities in the aluminum production, and cannot thus be included in a quantitative review of workers of carbon electrodes. In seven cohort studies on workers of carbon electrode manufacture, 97 cancers of the lung were observed versus 101.6 expected (pooled RR = 1.00, 95% CI 0.82 1.23, Table 8); for all respiratory cancers, 126 cancers were reported versus 135.7 expected (pooled RR = 0.97, 95% CI 0.81 1.15, Table 8 and Figure 7). Observed deaths were 16 for bladder cancer versus 13.0 expected (pooled RR = 1.35, 95% CI 0.83 2.20) and 20 versus 17.3 (pooled RR = 1.29, 95% CI 0.85 1.95) for all urinary tract cancers (Table 8). chimney sweeps Relatively few epidemiological studies have been conducted to quantify cancer risk among chimney sweeps. The largest 442 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology review Figure 6. Relative risks (RR) of bladder cancer among iron and steel foundry workers, and corresponding confidence intervals (CI), by study and overall. Figure 7. Relative risks (RR) of lung cancer (including other respiratory cancers not specified) among carbon electrode workers, and corresponding confidence intervals (CI), by study and overall. Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 443

cohort from Sweden [75], as well as a few other small cohorts, reported an excess risk from lung cancer [76 78]. No cohort study on chimney sweeps has been published since the review by Boffetta et al. [8]. conclusions The epidemiological data available on aluminum production workers, including more than 650 cancers of the lung and approximately 200 cancers of the bladder, indicate that there is no excess risk for cancer of the lung, but a modest excess risk for cancer of the bladder. The cause effect inference, however, remains unclear, giving the small excess risk and the limited data on dose- and duration-risk relation [79]. There is definite epidemiological evidence that coal gasification workers had high risks of respiratory and probably urinary neoplasms. Although these workers were mainly exposed to PAHs from coal tar, they may have been exposed to other contaminants, including heavy metals, silica and aromatic amines. Moreover, it is difficult to extrapolate indications from this kind of highly toxic production until 30 40 years ago. Data also indicate an excess risk for lung cancer, but not for other neoplasms, in workers of coke production, mainly of coke oven. For this group of workers, there are also quantitative exposure-risk estimates, which suggest a dose response relation with the cumulative dose of PAHs [1, 40]. A meta-analyses of 10 studies estimated a RR of 1.17 for an a priori defined unit of 100 lg/m 3 year of benzo[a]pyrene [80]. However, the comparability and validity of unit exposure measurements in various studies and populations remain open to discussion. In fact, only for a few cohorts a direct estimate of exposure was available. This criticism also applies to other industries and occupations considered in that meta-analysis, for which data are in any case more limited and inadequate for any inference of dose- and duration-relations. Iron and steel foundry workers had an excess risk of cancers of the respiratory tract, particularly until the early 1970s, with a significant dose risk relation [81]. Those workers, however, were exposed to various other known or potential carcinogenic substances, including several heavy metals, crystalline silica and asbestos. Therefore, it is questionable if the increased risk for respiratory cancers observed in these industries can be related to PAH exposure. The risk of bladder cancer was significantly increased in some studies, but there is no consistent epidemiological evidence on an excess risk. Data on workers exposed to bitumen fumes, coal tar and related products are limited and do not suggest any increased risk for both respiratory and urinary tract cancers, with the exception of a modest excess risk for lung cancer in roofers. Similarly, data are too limited to draw any conclusion on the association between exposure to carbon black and cancer of the respiratory and urinary tract. Along this line, Monograph no. 93 of the IARC concluded that the epidemiological studies of carbon black provided inadequate evidence of carcinogenity [82, 83]. No excess risk was observed for cancer of the respiratory and urinary system among carbon electrode workers [84]. For some cancer sites and industries, including mainly lung and respiratory tract cancer for coal gasification and coke production workers, the results from various studies were significantly heterogeneous. This probably reflects variable exposure patterns across different cohort and time periods, and indicates that due caution is required to generalize these estimates. Thus, the epidemiological evidence available to date confirms the existence of some excess risk for cancer of the lung in several PAH-related industries, with the exception of the aluminum and carbon electrode manufactures. The evidence for cancer of the bladder and of the urinary system is less consistent, with a modest increase in risk only for workers of aluminum production, coal gasification, iron and steel foundries. The excess risks observed were, however, modest for most industries (apart from coal gasification, which had an over two-fold increased risk), and whether such excesses are real or due at least partially to some bias or confounding cannot be completely ruled out. In particular, workers from the industries and occupations considered may have been exposed to other occupational carcinogens, as well as to tobacco smoking, which was not allowed for in most of the cohort studies considered. acknowledgements This work was conducted with the contribution of the Italian Association for Cancer Research and the Italian League Against Cancer. The authors thank Mr M. Plummer from IARC for providing the R function to plot the graphs, and Mrs I. Garimoldi for editorial assistance. references Annals of Oncology 1. Bostrom CE, Gerde P, Hanberg A et al. Cancer risk assessment, indicators, and guidelines for polycyclic aromatic hydrocarbons in the ambient air. Environ Health Perspect 2002; 110: 451 488. 2. International Agency for Research on Cancer. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 32: Polynuclear aromatic compounds, Part 1, Chemical environmental and experimental data. Lyon: International Agency for Research on Cancer 1984. 3. International Agency for Research on Cancer. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 33: Polynuclear aromatic compounds, Part 2, Carbon blacks, mineral oils and some nitroarenes. Lyon: International Agency for Research on Cancer 1984. 4. International Agency for Research on Cancer. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 34: Polynuclear aromatic compounds, Part 3, Industrial exposures in aluminum production, coal gasification, coke production, and iron and steel founding. Lyon: International Agency for Research on Cancer 1984. 5. International Agency for Research on Cancer. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 35: Polynuclear aromatic compounds, Part 4, Bitumens, coal-tars and derived products, shaleoils and soots. Lyon: International Agency for Research on Cancer 1985. 6. International Agency for Research on Cancer. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, suppl. 7: Overall evaluations of carcinogenicity: An updating of IARC monographs Volumes 1 to 42. Lyon: International Agency for Research on Cancer 1987. 444 Bosetti et al. Volume 18 No. 3 March 2007

Annals of Oncology 7. Straif K, Baan R, Grosse Y et al. Carcinogenicity of polycyclic aromatic hydrocarbons. Lancet Oncol 2005; 6: 931 932. 8. Boffetta P, Jourenkova N, Gustavsson P. Cancer risk from occupational and environmental exposure to polycyclic aromatic hydrocarbons. Cancer Causes Control 1997; 8: 444 472. 9. Boffetta P, Silverman DT. A meta-analysis of bladder cancer and diesel exhaust exposure. Epidemiology 2001; 12: 125 130. 10. International Agency for Research on Cancer. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 46: Diesel and gasoline engine exhausts and some nitroarenes. Lyon: International Agency for Research on Cancer 1989. 11. Greenland S. Quantitative methods in the review of epidemiologic literature. Epidemiol Rev 1987; 9: 1 30. 12. Milham S Jr. Mortality in aluminum reduction plant workers. J Occup Med 1979; 21: 475 480. 13. Giovanazzi A, D Andrea F. Cause di morte fra i lavoratori di un impianto di riduzione elettrolitica di allumina. Med Lav 1981; 4: 277 282. 14. Andersen A, Dahlberg BE, Magnus K, Wannag A. Risk of cancer in the Norwegian aluminum industry. Int J Cancer 1982; 29: 295 298. 15. Ronneberg A, Andersen A. Mortality and cancer morbidity in workers from an aluminum smelter with prebaked carbon anodes Part II: Cancer morbidity. Occup Environ Med 1995; 52: 250 254. 16. Ronneberg A, Haldorsen T, Romundstad P, Andersen A. Occupational exposure and cancer incidence among workers from an aluminum smelter in western Norway. Scand J Work Environ Health 1999; 25: 207 214. 17. Romundstad P, Haldorsen T, Andersen A. Cancer incidence and cause specific mortality among workers in two Norwegian aluminum reduction plants. Am J Ind Med 2000; 37: 175 183. 18. Romundstad P, Haldorsen T, Andersen A. Lung and bladder cancer among workers in a Norwegian aluminum reduction plant. Occup Environ Med 2000; 57: 495 499. 19. Romundstad P, Andersen A, Haldorsen T. Cancer incidence among workers in six Norwegian aluminum plants. Scand J Work Environ Health 2000; 26: 461 469. 20. Rockette HE, Arena VC. Mortality studies of aluminum reduction plant workers: potroom and carbon department. J Occup Med 1983; 25: 549 557. 21. Gibbs GW. Mortality of aluminum reduction plant workers, 1950 through 1977. J Occup Med 1985; 21: 761 770. 22. Mur JM, Moulin JJ, Meyer-Bisch C et al. Mortality of aluminum reduction plant workers in France. Int J Epidemiol 1987; 16: 257 264. 23. Moulin JJ, Clavel T, Buclez B, Laffitte-Rigaud G. A mortality study among workers in a French aluminum reduction plant. Int Arch Occup Environ Health 2000; 73: 323 330. 24. Spinelli JJ, Band PR, Svirchev LM, Gallagher RP. Mortality and cancer incidence in aluminum reduction plant workers. J Occup Med 1991; 33: 1150 1155. 25. Carta P, Cocco PL, Fiore C et al. Mortalità nei lavoratori della fonderia di alluminio primario di Portovesme in Sardegna. Med Lav 1992; 83: 530 535. 26. Carta P, Aru G, Cadeddu C et al. Mortality for pancreatic cancer among aluminum smelter workers in Sardinia, Italy. G Ital Med Lav Ergon 2004; 26: 83 89. 27. Doll R, Vessey MP, Beasley RW et al. Mortality of gas-workers final report of a prospective study. Br J Ind Med 1972; 29: 394 406. 28. Manz A. Atem- und Harnwege als Lokalisationsstellen berufsbedingter (Teer-) Karzinome bei Kokerei- und Rohrnetzarbeitern. VDI-Berichte 1980; 358: 227 235. 29. Hansen KS, Viskum S, Pedersen MS. Dødelighed blandt gasvaerksarbejdere. Ugeskr Laeger 1986; 148: 610 612. 30. Gustavsson P, Reuterwall C. Mortality and incidence of cancer among Swedish gas workers. Br J Ind Med 1990; 47: 169 174. 31. Berger J, Manz A. Cancer of the stomach and the colon-rectum among workers in a coke gas plant. Am J Ind Med 1992; 22: 825 834. 32. Reid DD, Buck C. Cancer in coking plant workers. Br J Ind Med 1956; 13: 265 269. 33. Sakabe H, Tsuchiya K, Takekura N et al. Lung cancer among coke oven workers: a report to Labour Standard Bureau, Ministry of Labour, Japan. Ind Health 1975; 13: 57 68. review 34. Davies GM. A mortality study of coke oven workers in two south Wales integrated steelworks. Br J Ind Med 1977; 34: 291 297. 35. Hurley JF, Archibald RMcL, Collings PL et al. The mortality of coke workers in Britain. Am J Ind Med 1983; 4: 691 704. 36. Wu W. Occupational cancer epidemiology in the People s Republic of China. J Occup Med 1988; 30: 968 974. 37. Swaen GMH, Slangen JJM, Volovics A et al. Mortality of coke plant workers in the Netherlands. Br J Ind Med 1991; 48: 130 135. 38. Franco F, Chellini E, Seniori Costantini A et al. Mortality in the coke oven plant of Carrara, Italy. Med Lav 1993; 84: 443 447. 39. Chau N, Bertrand JP, Mur JM et al. Mortality in retired coke oven plant workers. Br J Ind Med 1993; 50: 127 135. 40. Costantino JP, Redmond CK, Bearden A. Occupationally related cancer risk among coke oven workers: 30 years of follow-up. J Occup Environ Med 1995; 37: 597 604. 41. Bye T, Romundstad PR, Ronneberg A, Hilt B. Health survey of former workers in a Norwegian coke plant: Part 2. Cancer incidence and cause specific mortality. Occup Environ Med 1998; 55: 622 626. 42. Koskela RS, Hernberg S, Kärävä R et al. A mortality study of foundry workers. Scand J Work Environ Health 1976; 2: 73 89. 43. Gibson ES, Martin RH, Lockington JN. Lung cancer mortality in a steel foundry. J Occup Med 1977; 19: 807 812. 44. Breslin P. Mortality among foundry men in steel mills. In Lemen R, Dement JM (eds): Dusts and Disease. Park Forest South, IL: Pathotox Publishers 1979; 439 447. 45. Decouflé P, Wood DJ. Mortality patterns among workers in a gray iron foundry. Am J Epidemiol 1979; 109: 667 675. 46. Andjelkovich DA, Mathew RM, Richardson RB, Levine RJ. Mortality of iron foundry workers: I. Overall findings. J Occup Med 1990; 32: 529 540. 47. Andjelkovich DA, Mathew RM, Yu RC et al. Mortality of iron foundry workers: II. Analysis by work area. J Occup Med 1992; 34: 391 401. 48. Andjelkovich DA, Shy CM, Brown MH et al. Mortality of iron foundry workers: III. Lung cancer case-control study. J Occup Med 1994; 36: 1301 1309. 49. Hansen ES. Cancer mortality among Danish molders. Am J Ind Med 1991; 20: 401 409. 50. Sherson D, Svane O, Lynge E. Cancer incidence among foundry workers in Denmark. Arch Environ Health 1991; 46: 75 81. 51. Rotimi C, Austin H, Delzell E et al. Retrospective follow-up study of foundry and engine plant workers. Am J Ind Med 1993; 24: 485 498. 52. Sorahan T, Faux AM, Cooke MA. Mortality among a cohort of United Kingdom steel foundry workers with special reference to cancers of the stomach and lung, 1946 90. Occup Enviran Med 1994; 51: 316 322. 53. Moulin JJ, Clavel T, Roy D et al. Risk of lung cancer in workers producing stainless steel and met allic alloys. Int Arch Occup Environ Health 2000; 73: 171 180. 54. Maclaren WM, Hurley JF. Mortality of tar distillation workers. Scand J Work Environ Health 1987; 13: 404 411. 55. Moulin JJ, Mur JM, WiId P et al. Etude épidémiologique de mortalité parmi les salariés d une usine de distillation des goudrons de houille. Rev Epidém Santé Publ 1988; 36: 99 107. 56. Swaen GM, Slangen JM. Mortality in a group of tar distillery workers and roofers. Int Arch Occup Environ Health 1997; 70: 133 137. 57. Bogovski P. Historical perspectives of occupational cancer. In Vainio H, Sorsa M, Hemminki K (eds): Occupational Cancer and Carcinogenesis. Washington DC: Hemisphere 1981; 921 939. 58. Miller BG, Cowie HA, Middleton WG, Seaton A. Epidemiologic studies of Scottish oil shale workers: III. Causes of death. Am J Ind Med 1986; 9: 433 446. 59. Karlehagen S, Andersen A, Ohlson CG. Cancer incidence among creosoteexposed workers. Scand J Work Environ Health 1992; 18: 26 29. 60. Wong O, Harris F. Retrospective cohort mortality study and nested case-control study of workers exposed to creosote at 11 wood-treating plants in the United States. J Occup Environ Med 2005; 47: 683 697. 61. Hammond EC, Selikoff IJ, Lawther PL, Seidman H. Inhalation of benzpyrene and cancer in man. Ann NY Acad Sci 1976; 271: 116 124. Volume 18 No. 3 March 2007 doi:10.1093/annonc/mdl172 445