Cancer incidence in an occupational cohort exposed to bitumen fumes

HANSEN ES. Cancer incidence in an occupational cohort exposed to bitumen fumes. Scand J Work En viron Health 1989;15:101-5. This study was conducted to investigate whether bitumen fumes should be considered carcinogenic to human beings. A historical cohort of heavily exposed mastic asphalt work ers was followed from 1959through 1984(inclusive) with regard to cancer incidence. A total of 679 Dan ish men were included in the study cohort. Among these, 75 new cases of cancer were observed within the period studied . The cancer incidence observed among the group significantly exceeded that of the total Danish male population, the standardized morbidity ratio (SMR) being 195 (95 % confidence interval (95 "70 CI) 153-244). Significant increases were seen for cancer of the mouth (SMR 1111,95 % CI 135-4014), the esophagus (SMR 698, 95 % C1 144-2039), the rectum (SMR 318, 95 % CI 128-656), and the lung (SMR 344, 95 % CI 227-501). It is suggested that exposure to cracking products in the fumes of heated bitumen has contributed to the elevated cancer incidence observed.

Bitumens a re produced as residues from the fractional distillation of crude oil. They are widely used becau se of the ir adhesive and waterproofing properties. It has been estimated that the current annual world use is over 60 million tons (I). Major applications are in pa ving for roads and airfields, hydr aulic uses (such as dams , water reservoirs, and sea-defense works ), roofing, flooring, and the protection of metal s again st cor ros ion . Bitumens contain onl y relati vely small amounts of pol ycyclic aromatic hydrocarbons (P A H) (l ). [The rea son for these low amounts ma y be that the refiner y proce ss used for the manufactur ing of bitumen s remo ves the majority of compounds with lower boiling point s, includ ing PAH with three to seven fused rings , which include all of the PAH compounds which have thu s far been con sidered to have carcinogenic effect s by the international Agency for Research on Cancer (2)]. The recommended temperatures for handling bitumen-based asphalt mixes range from 65 to 230°C, depending on the type of bitumen (I). The generation of bitumen fumes increases steeply as the temperature increases (3). Besides, the heating may initiate cracking processes which may cause substances not present in the unheated bitumen to form. Several countries, including Denmark , have adopted a timeweighted average (TWA) of 5 mg/m' as the administrative limit for occupational expo sure to bitumen fum es. P revious epidemioiogic studies ha ve shown an increased cancer mortality among roofers, who se occup ati onal expo sure includes heated bitumen and bit-I Institute of Commu nity Hea lth, University of Odense, Odense, Denmar k.
2 umen fume s. Men ck & Henderson (4) used registry data to estim ate the lung ca ncer mortality for a number of occupational gro ups in Los Angeles County in the United State s. For roofers, the y found a lung cancer mortality that was about five time s that of the population at large . Another study on roofers ' mortality was carr ied out by Hammond and his co-workers (5). The y conducted a 12-year follow-up of a cohort of roofers identified from union files in the United States. Am ong long-term employed roofers, they found an increased mortality from malignant neoplasms in the buccal cavity and pharynx , the d igestive organs, the respiratory system, the pro state gland , the bladder , and the skin and from leukem ia . Howe ver, although the stud ied groups of roofers (4, 5) had been exposed to bitumen fumes, no clear conclu sions could be dra wn regarding the carcinogenicity of the se fumes becau se the roofers had been concomitantly expo sed to the known car cinogen coal tar. Only a few experimental studies have been carried out on the potential carcinogenic effects of bitumen fumes. Inhalation studies on rats and guine a pigs (6) and on mice (7) have failed to demonstrate any carcinogenic effect of bitumen fumes. Nor could an effect be demonstr at ed in a po sitive control group exposed to coal-tar fume s (6). The employed animal exposure model mu st thus be rather insensitive, and the apparently negati ve findings are not info rmative regardin g the potential carcinogenic effects of bitumen fume s. However, a recent animal assay using skin application has shown a strong carcinogenic effect of conden sate o f bitumen fume s on mice (8). Sixty-six percent of the anim als developed malignant neoplasms after skin paintin g with condensated bitumen fumes . The effect of bitumen fume condensate was almost as strong as th at seen in the positi ve contro l gro up, in which the animals were painted with cond ensated coal-tar fumes derived at the same temperatures as the bitumen fumes. The concentration of known carcinogenic P AH was much lower in the condensate of bitumen fumes than in the coal-tar fumes. Chemicals other than the measured P AH must therefore be involved in producing the observed strong carcinogenic effects of bitumen fumes, either as initiators, promotors, or as cocarcinogens.
The present study was set up to investigate further the relationship between human cancer risk and exposure to bitumen fumes. The study considered the cancer incidence in an occupational cohort of Danish mastic asphalt workers heavily exposed to bitumen fumes only.

Subjects and methods
A historical cohort of male mastic asphalt workers was set up from employment lists and union records and followed with regard to cancer incidence during the period 1959-1984.

Exposure
Mastic asphalt is a mixture of fine sand, stone powder, and finely divided limestone with a high content (12-17 010) of hard bitumen. Mastic asphalt is used for surfacing roads and in flooring and roofing. The asphalt mix is manufactured at an asphalt plant, poured into heated, closed tanks, and carried to the place where it is to be applied. The asphalt mixture is kept at a temperature of approximately 250°C until it is filled into buckets. The asphalt workers carry the buckets containing the hot asphalt mixture to the place of application, where it is tipped out and layered by hand floating -a leveling technique similar to that used for smoothing cement. On his knees, the float troweler draws out the hot asphalt mixture with a long wooden trowel. While handling the hot mastic asphalt mixture, the workers may be exposed to high concentrations of bitumen fumes. Hygienic measurements carried out by the Danish National Institute of Occupational Health (Rietz B, internal report of the Danish National Institute of Occupational Health, in Danish) indicate that the TWA of 5 mg/m' is being considera- bly exceeded during flooring with mastic asphalt. A total of 35 personal samples was taken during flooring. The concentration of asphalt fume condensate in these samples ranged from 0.5 to 260 mg/m' with a median of 19.7 mg/m' and a mean of 41.5 mg/m'. The amount of P AH was determined in nine of the 35 samples for an estimation of the PAH concentration in the workplace air. For total P AH (16 standards), the median equaled 0.183 mg/rn", and the mean was 0.195 mg/m'. For benzo(a)pyrene, the median was 0.004 mg/m', and the mean 0.0058 mg/m'. Only two samples were taken during manual road paving. The concentration of asphalt fume condensate was estimated to be 4.3 mg/m' in one and 3.4 mg/m' in the other. Results similar to these have been reported in several other studies on mastic asphalt workers' exposure to asphalt fumes during flooring and paving.
(See reference 1.) As mentioned previously, mastic asphalt is a mixture of bitumen and mineral aggregates. The mastic asphalt mixture contains an excess of bitumen, and the mineral aggregate used is very fine-grained, which ensures that the technical properties of the mixture are optimal. Therefore, technical auxiliary substances such as coal tar are not added to the mastic asphalt mixture. However, during World War II a shortage of bitumen initiated the use of coal-tar pitch in the production of asphalt mixtures, including that of mastic asphalt.

Identification and follow-up of the subjects
A total of 679 mastic asphalt workers, all men, was entered into the study. The subjects were identified from historical files that covered the period 1959 through 1980. Four hundred subjects were identified from the employment lists of four mastic asphalt plants, 186 subjects from the membership files of an organized group of mastic asphalt workers affiliated with the National Union of General Workers, and 93 subjects from the membership files of a benefit society organized by the workers at one of the mastic asphalt plants. A subject was identified either by name and address, name and date of birth, or by his personal identification number and enrolled into the study by the time when he was first identified in one of the historical files used for establishing the study cohort. The follow-up of a subject was stopped at death, emigration, or on I January 1985, whichever occurred first. Information on death and emigration was obtained from the local and central offices of the national register, and by this procedure everyone was traced. By the end of the follow-up, ie, on I January 1985, 524 subjects were alive and living in Denmark, 149 had died, and 6 had emigrated. The study comprised 6 692 person-years at risk (table 1). The occurrence of new cases of cancer was ascertained through a search for the cohort members in the Danish Cancer Register. Table 2. Cancer incidence, standardized morbid ity ratio (SMR), and 95 % con fid ence inte rval for the 6 79 masti c asphal t workers, all ages combi ned . Table 3. Site-spec ifi c ca nce r incidence, st andardized morb id ity rati o (SMR), and 95 % confi dence int erval for th e 547 mast ic asphalt work ers aged 40-89 years.

Data analysis
The expe cted ca ncer incidence in the cohort of mastic asphalt workers was calculated from age-, period-, and site-specific cancer incidenc e rates for Dani sh men during the period 1958-1982. Becau se data on each individual's expo sure histo ry were generally not ava ilable, ind uction/ latency time from first exposure to the establishment of the cancer diagnosis has not been considered in th e analysis. However, the age gro uping represents a proxy for an analysis by time from first exposure becau se new ma stic asphalt workers ar e always young, typically in th eir early twenties, when they sta rt this job and becaus e the turnover is onl y mod erate in thi s occup ational gro up. Thus, th e ana lyses of ma stic asphalt workers aged 40 years or more may be considered as approximatel y equal to anal yses in which a 15to 20-year inductio n/ latency time is required. Albeit a crude prox y, anal ysis by age seemed to be the onl y alternative to ignoring the induction/ latency time totally.
The study cohort included persons born wit hin the years 1893-1960. It is therefor e likely to be rather inhomogeneous as regards several risk factors for cancer. In addition , some of the subjects ma y ha ve been asphalt workers dur ing World War II and may have been expo sed to coal-tar pitch. For the se rea son s, the cohort of ma stic asphalt wor kers was divided into three subcohor ts by th e subj ects' time of birth as fo llows : subcoho rt I: persons bo rn in 1893-1919 (coal-ta r exposure during World War II likely), subcohor t II : persons born in 1920-1929 (some coal-tar expo sure du ring World War II possible), and subcohort III : perso ns born in 1930-1960 (coal-tar exposure during World War II impossib le).
The calculation o f the tests and co nfidence intervals was based on the ass ump tion that the ob ser ved number s were P oisson -distributed .

Results
The observed cancer incide nce am ong th e ma stic asphalt wo rkers was almost two tim es grea ter than tha t o f the total population . The increased cancer incidence deri ved mainly from malignancies of the respiratory system and the digesti ve organs (table 2). For persons aged 40 years or mo re , statistically significant increases were seen for cancer of the mouth, esophagus, rectum, and lung (ta ble 3). Twenty-seven cases of primary lung ca ncer were o bserved am on g the mastic aspha lt workers (ta ble 4). Of th e 27 lun g cancer cases, 18 occurred in subcohort I , six in subcohort II, and three in subcohort III . The lung cancer incidence was significantly increased in each of the three subcohorts (table 5).
As for ot her diagnoses, the result s for the subcohorts are not shown becau se almost all of th e cases occurred among workers aged 65 years or ove r, and there were no men of these ages in subcohorts II and III.

Discussio n
Th is study showed an increased cancer incidence in the cohort of mastic asphalt work ers. An obv ious explanation may be that hot ma stic asphalt involves exposure to carcinogenic substances. But before drawing thi s conclusion , one needs to con sider the potential bia s in the stu dy.
It was not possible to control for potential differences in life-style (including smok ing, drinking, and dietary habits) between the cohort of mastic asphalt workers and the Danish men of the same age in general dur ing the same period. In particular, control for smoking habits would have been advantageous, because smoking is intimately associated with respiratory cancer. Smoking, however, is so preva lent amo ng Danish men that , even if all the asphalt workers had been heavy smokers, smoking could expla in no mor e than a doubling of their lung cancer risk as compared with that of the general population (9). An inquiry in 1976 to mastic asphalt workers in Copenhagen showed that 22 % were nonsmokers, 36 % were medium smokers (ie, daily consumption 1-14 g of tobacco), and 43 % were heavy smokers (ie, dail y consumption 15 g of tobacco or more) (10). Similarly, a population sur vey in 1982 showed that, among men of the same age, 39 % were nonsmokers, 24 % were medium smokers, and 38 % were hea vy smokers (II) . Smoking differences of this order of magnitude will bring about approximately 20 % more lung cancer cases than " expected " in the group of mastic asphalt workers . The observed lung cancer incidence, however, was more than three times greater than expected . Therefore, though smoking present s an impo rtant risk of lung cancer, it is hardly likely to pla ya major role as a confounding factor in the present study.

104
The incidence rates for respiratory and digestive cancer among Danish men shows a rural-urban gradient (12). Almost all of the asphalt workers were city dwellers as opposed to only 40 % of the comparison population (13). Therefore, the expected incidence of respi ratory cancer may ha ve been underestimated by 35 %, where as the expected incidence of digestive cancer ma y have been underestimated by 10 % .
After adju stment for the estimated confounding effect of urbanization and smoking habits, the observed incidence of both respiratory and digestive cancer was about two times greater than expected (adjusted standardized morbidity ratio for respiratory cancer 214 and for digestive cancer 206). All of the 679 mastic asphalt worke rs considered in this stud y had been expo sed to asphalt fumes , but only some of them had been exposed long-term. Data on ind ividual exposure history are lacking, but a previous quest ionnaire study among the mastic asphalt workers showed th at half of them had been employed for less than 10 years and one-third for less than six years (10). Furthermore, it is necessary to consider the fact that mastic asphalt workers have been casual labo rers and typically out of work during the winter months. As regards exposure intensity, the concentration of asphalt fume s has undoubtedly been much higher during flooring ope rations tha n during road paving . Roughly estimated, the concentration has been a bout 4 mg/m ' d uring road paving and about 20 mg/m' during flooring operations (Rietz B, internal report of the Danish National Institute of Occupational Health, in Da nish). Road paving has made up approximately two-thirds and flooring operations one-third of the workhours of the cohort in question. Thus, weighted over a 12-month per iod, the mastic asphalt workers' average expo sure has been almost identical to that of continuous exposure to asphalt fume s at the current Danish TWA of 5 mg/m' . The subcohort analysis revealed that the increased lung cancer risk had not been restricted to the mastic asphalt workers who had possibly been exposed to coa ltar pitch during World War II. As a matter of fact , th e increa sed lung cancer risk tended to be manifest earlier (ie, at a younger age) in the subsequent birth cohorts. The numbers are too small, however, to justi-fy any conclusions as to a possible trend of increasing risk in the subsequent birth cohorts.
In conclusion, the results of this study indicate that the occupational exposure of mastic asphalt workers is associated with an increased risk of malignant neoplasms in the buccal cavity, the respiratory system, and the digestive organs. These results are in agreement with those of studies on roofers' mortality (4,5). The occupational exposure of roofers resembles that of mastic asphalt workers, as both groups are exposed to fumes of heated bitumen. Bitumen fume is a disperse aerosol, the particles of which contain many different hydrocarbons. If this aerosol is inhaled, the small particles are deposited mainly in the deep, nonciliated airways, whereas the larger particles are deposited on the ciliated epithelium of the conducting airways. The latter particles will be cleared from the airways by the mucociliary escalator mechanism and subsequently swallowed. This way, even the digestive organs may be exposed to the substances contained in bitumen fumes. Soiling of hands, face, and workclothes also presents a risk of exposure via the mouth. The fumes from hot mastic asphalt contain the most volatile substances of the actual bitumen in addition to potential cracking products. The content of carcinogenic PAH is low in bitumen fume, as compared with that of heated coal tar (I, 8). Despite this difference, bitumen fumes have been shown to produce almost as many experimental cancers as have coal-tar fumes (8). These findings, combined with the fact that unheated bitumen is a rather ineffective carcinogen (I), point to the possibility that carcinogenic substances in bitumen fume may originate from cracking processes initiated by the heating of bitumen. Thus, in reference to the frame for causal inference which has been proposed by Hill (14), there seems to be reason to believe the hypothesis that bitumen fumes are carcinogenic.
The observed association is too strong to be explained by confounding or by random variation, and it has been demonstrated in different populations. Furthermore, the epidemiologic findings are biologically plausible, and the hypothesis is supported by experimental evidence. As to possible weak points in this reasoning, it cannot be totally excluded that the epidemiologic findings may have been brought about by another carcinogenic agent in the work environment. However, it is not likely that the effects demonstrated in the present study should be due solely to an unnoticed occupational risk factor.