Diesel engine exhaust and lung cancer: an unproven association.

The risk of lung cancer associated with diesel exhaust has been calculated from 14 case-control or cohort studies. We evaluated the findings from these studies to determine whether there is sufficient evidence to implicate diesel exhaust as a human lung carcinogen. Four studies found increased risks associated with long-term exposure, although two of the four studies were based on the same cohort of railroad workers. Six studies were inconclusive due to missing information on smoking habits, internal inconsistencies, or inadequate characterization of diesel exposure. Four studies found no statistically significant associations. It can be concluded that short-term exposure to diesel engine exhaust (< 20 years) does not have a causative role in human lung cancer. There is statistical but not causal evidence that long-term exposure to diesel exhaust (> 20 years) increases the risk of lung cancer for locomotive engineers, brakemen, and diesel engine mechanics. There is inconsistent evidence on the effects of long-term exposure to diesel exhaust in the trucking industry. There is no evidence for a joint effect of diesel exhaust and cigarette smoking on lung cancer risk. Using common criteria for determining causal associations, the epidemiologic evidence is insufficient to establish diesel engine exhaust as a human lung carcinogen.

cohort studies. We evaud the fi fom these udies to d whether dhre is sufficient idence t ica it a h hug igudies fund increased risks associaftewnit two of th:isourst.adI wete base on the hort of ad woe. Six stui e a use d i in on -smokinghabits, inter U, or i; a odiese Foulr studies find no statistical n be ioded tat short-rm exposure to diese engine esxhust (40 yea) does not ha a cauative role-in human lung cancer. There is but not.:cnsa evidence that long-term cure to diel exhaust (>20 years) s the risk of lun-mite.r 1cmoi en , as en, and diese engine mechanics. Thre is in tt e on the effc of ltterm poue o diesel ehust Diesel engine exhaust contains respirable carbonaceous particulates that adsorb organic chemicals, including the polycyclic aromatic hydrocarbons benzo[a]pyrene and 1-nitropyrene. These organic extracts are carcinogenic to rodents when administered topically or by implantation. Inhalation of high concentrations of whole diesel exhaust causes destruction of defensive pulmonary mechanisms and promotes the development of primitive lung adenocarcinoma in animal models (1). At lower levels of exposure that do not reduce pulmonary clearance, diesel exhaust is not carcinogenic. This suggests that the mechanism of carcinogenic action for diesel inhalation is particle overloading and subsequent inflammation of the lung, and not the mutagenic effects of the organic fraction of diesel exhaust (2,3). The results from recent experiments support this concept. Many rats were chronically exposed to high levels of diesel exhaust or carbon black (4). Carbon black has a similar carbonaceous core as diesel exhaust minus the mutagenic organic fraction. Both diesel exhaust and carbon black produced the same incidence of lung tumors in experimental rats. The relevance of these laboratory findings has possible implications for interpreting studies of diesel exhaust and human lung cancer. Cigarette smoke particulates cannot induce damage to human bronchial epithelium when the lung's ciliated mucusproducing epithelium is intact (5). It is therefore uncertain whether the much lower concentrations of diesel exhaust in traffic or industrial settings, compared with animal inhalation studies, can cause bronchial damage and subsequent cancer in humans. Furthermore, unlike cigarette smoke inhaled through the mouth, diesel exhaust is inhaled through the nose and encounters upper respiratory defense mechanisms.
Elevated mortality ratios of lung cancer have been documented in industries that use diesel engines. These observations have implicated diesel exhaust as a possible human lung carcinogen. In recent years, several case-control and cohort studies have been conducted to assess the independent role of diesel emissions in lung cancer. This paper reviews the epidemiologic findings and evaluates the evidence for causality according to standard criteria. Prior Reviews Diesel engines have been used increasingly in various industries since the 1930s, although they were not in widespread use until about 1950. Diesel engines are the power source of railroad locomotives, heavy equipment vehicles, and some buses and trucks. Diesel engines are also used in mining and dock operations. The health effects of diesel exhaust have been critically addressed by several groups. In 1981, the National Research Council (NRC) of the National Academy of Sciences found no evidence for a carcinogenic effect of diesel exhaust in epidemiologic studies, although the lack of high-quality research in this area was acknowledged (6). I.T.T. Higgins, a member of the NRC committee, stated in a separate position paper that the cancer risk from diesel exhaust emissions was "uncertain" (). Concerns were raised about inadequate allowance for asbestos exposure and cigarette smoking in occupational cohort studies. A similar conclusion was reached by Wynder and Higgins in 1986 (8). Reviews by Schenker (9) and Steenland (10) concluded that the evidence was suggestive but inconclusive. The International Agency for Research on Cancer concluded in 1989 that based on the evidence from animal studies, "diesel engine exhaust is probably carcinogenic to humans" (11). The U.S. Environmental Protection Agency has proposed classifying diesel exhaust as a probable human carcinogen (12). The National Institute for Occupational Safety and Health concluded that diesel exhaust is a potential. human carcinogen (13).
These evaluations were based primarily on results of experimental animal studies and elevated standardized mortality ratio statistics of lung cancer in some dieselexposed workers. The epidemiologic studies were done on truckers and other motor vehicle drivers (14-28), traffic controllers (29), coal miners (30), construction workers (18,25,31,32), railroad union members (16,33) and dock workers (34). These investigations lacked accurate exposure information on diesel exhaust and individual smoking habits or had insufficient follow-up times to account for the potential latent effects of diesel exposure.
More recent case-control and cohort studies have provided additional information on the health effects of diesel exhaust (Table 1). These studies are reviewed here.

Case-Control Studies of Lung Cancer
In a French case-control study conducted by Benhamou et al. (35), the smokingadjusted odds ratio (OR) was 1.42 (95% CI, 1.07-1.89) for motor vehicle drivers. The types of motor vehicles were not specified. There was no trend in the odds ratio with duration of employment. Information on exposure to diesel exhaust was not obtained.
Damber et al. (36) interviewed surro-  (37) interviewed 502 male case-control pairs in hospitals. Using industrial hygiene criteria to define diesel exposure, the crude OR was 2.0 (95% CI, 1.2-3.2). After adjustment for cigarette smoking, the OR was 1.4 (95% CI, 0.8-2.4). Using National Institute of Occupational Safety and Health criteria to define exposure, the unadjusted risk was 1.7 (95% CI, 0.6-4.6) for a high probability of exposure, and 0.7 (95% CI, 0.4-1.3) for a moderate degree of exposure. There were no specific occupational groups that had an elevated risk of lung cancer. The statistical analyses were based on the usual occupation. Information on duration of employment was not analyzed.
Siemiatycki et al. (38) conducted a population-based study of multiple cancer sites. They Interviewed 3726 male cancer patients in Montreal hospitals. An industrial hygienist classified each job according to the potential for exposure to diesel exhaust and other emissions or air pollutants. The risk of squamous cell lung cancer for "non-substantial" diesel exhaust was 1.9 (90% CI, 1.0-3.5) and "substantial" (defined as exposure levels above the median cumulative exposure index) diesel exhaust was 1.2 (90% CI, 0.6-2.4) compared with unexposed subjects. Compared to subjects exposed to "nonsubstantial" gasoline exhaust, the respective risk estimates were 2.3 (90% CI, 1.0-5.2) and 1 The risk for diesel-exposed truck drivers who worked 1-15 years compared to truck drivers who were unexposed to diesel was 0.87 (95% CI, 0.33-2.25).
The RR for heavy equipment operators was 2.6 (95% C, 1.12-6.06), although this was based on only five lung cancer deaths. The RR for miners was 2.67 (95% CI, 1.63-4.37). The increased risk associated with mining should be interpreted with caution because it was based on 15 lung cancer deaths, and some mining operations have never used diesel-fueled motors in underground pits (41). In addition, high levels of radon daughters in some mines increase the risk of lung cancer (42). The evidence from this cohort study does not implicate diesel exhaust as a lung cancer risk factor.

Railroad Workers
Howe et al. (43) examined the lung cancer mortality experience of 43,826 retired Canadian railway workers employed from 1965 to 1977. The probability of exposure to diesel fumes was evaluated by the Department of Industrial Relations. The relative risk was 1.20 (p<0.013) for possible exposure, and 1.35 (p<0.001) for probable exposure. There were no data on years of employment and smoking habits. The mortality rates of other tobacco-related cancers (except bladder) and emphysema among diesel-exposed employees were slighdy elevated, suggesting that the smoking prevalence was higher than for other workers. The results from this well-done study require careful interpretation due to the lack of information on cigarette smoking, asbestos exposure, and duration of diesel exposure.
Garshick et al. (44) compared 1,256 lung cancer deaths to two age-matched controls in a retrospective cohort study of 650,000 active and retired railroad workers. The baseline study year was 1959, when diesel engines had nearly replaced all steam engines in the railroad industry (45).
Consequently, few workers were exposed to asbestos. Information on cigarette smoking habits was obtained from the next of kin. An industrial hygienist conducted sampling tests to detect the levels of diesel exhaust in selected jobs. The extent of diesel exposure in other job categories was determined by job activities and degree of contact with diesel equipment. This exposure classification system was verified against a survey sent to each worker in these jobs. Asbestos exposure was based on surveys and on medical and industrial literature, although exposure to asbestos occurred primarily during the steam-engine era. An increased risk of lung cancer was found for younger employees (<65 years of age) who worked in a diesel-related job for 20 or more years (OR = 1.41, 95% CI, 1.05-1.88). This risk was adjusted for pack-years of smoking and asbestos exposure.
The methodologic advantages of this study include a more precise exposure assessment of diesel exhaust and statistical adjustment for cigarette smoking (packyears) and asbestos. A sufficient latent period was allowed for, and data on long-term exposure were available. A possible bias was imprecise smoking histories obtained from next of kin. This study provides evidence of a risk associated with long-term diesel exhaust exposure. However, surrogate information on cigarette smoking is often inaccurate (46,4), and no association with lung cancer was found for older workers. The authors state that older workers were exposed to diesel exhaust for only a short period, although this needs verification.
The same group conducted a retrospective cohort study of 55,407 white, male railroad workers who were exposed to little or no asbestos (48). Members of the cohort had worked for 10-20 years in the railroads after 1959. Jobs with possible exposure to diesel were identified by job titles and job descriptions. An industrial hygiene survey was subsequently conducted to determine the probability of exposure in these jobs. Most railroad workers kept the same jobs during their tenure. Death certificates were obtained on 88% of the cohort. There were 1694 deaths attributed to lung cancer. For older employees who were exposed to diesel for less than 20 years as of 1980, there was no increased rate of lung cancer. The relative risk associated with 20 years of diesel exposure was 1.45 (95% CI, 1.11-1.89) for 40-44 year olds and 1.33 (95% CI, 1.03-1.73) for 45-49 year olds. These younger groups also had the lowest exposure to asbestos. There was no trend with increasing years of diesel exhaust, but when the most recent years of exposure were excluded from the analysis (the four years preceding death), a trend in cumulative exposure was found. Some concerns in this study include the ambiguous data on trend tests and the lack of information on cigarette smoking. However, it is likely that there was little variability in socioeconomic class and therefore the workers may have had similar smoking habits. It should be noted that the two studies of lung cancer by Garshick et al. (44,48) were conducted in the same occupational setting.
Motor Vehide Drivers and Mechanics Gustavsson et al. (49) examined the cancer incidence of 695 bus garage workers in Sweden between 1945 and 1970. These workers were employed as mechanics, servicemen, or hostlers for at least 6 months. All motorized buses in Sweden have been diesel-powered since the end of World War II. Twenty cases of lung cancer occurred in this group. The intensity of the exposure to occupational diesel exhaust and asbestos was assessed by industrial hygienists. The authors fitted a weighted regression model using a cumulative exposure measure to subjects with lung cancer and to subjects who died of lung cancer. The statistically significant relative risks were 1.34 for lowlevel exposure to diesel exhaust and 2.43 for high-level exposure to diesel exhaust. There were no data on smoking habits.
Steenland et al. (50) conducted a case-control study of Teamsters Union members. Death certificates were obtained for more than 10,000 members who filed claims for pension benefits. At least 20 years of tenure in the union was required to claim benefits. Of these, 1,288 men died of lung cancer. The Teamsters records did not have information on the types of truck engines used by members. Data were obtained on the number of years employed in each job. Additional information on occupation, smoking history, and asbestos exposure was obtained from next-of-kin.
Members were classified into jobs with potential diesel exposure based on job category (e.g., diesel truck driver, gasoline truck driver, etc.).
Using the Teamster employment data, elevated but nonsignificant odds ratios between 1.27 and 1.69 were found for long-haul drivers, short-haul drivers, truck mechanics, and other jobs with possible diesel exposure, after adjusting for smoking. The risk of lung cancer by duration of employment after 1959 (the approximate year when most trucks were diesel-powered in the United States) for long-haul truck drivers was, for 1-11 years, 1.08 (95% CI, 0.68-1.70), for 12-17 years, 1.41 (95% CI, 0.9-2.21), for >18 years, 1.55 (95% CI, 0.97-2.47) (linear trend test, p<0.05). No trend in duration was found for shorthaul truck drivers or for truck mechanics.
From the next-of-kin interviews, nonsignificant odds ratios between 1.25 and 1.54 were found for truck drivers and other jobs with potential diesel exposure. Diesel truck drivers who were employed for 1-24 years had no significantly increased risk of lung cancer. The odds ratio for 56 drivers who drove diesel trucks for 35 years or more was 1.89 (95% CI, 1.04-3.42) after adjusting for smoking. However, for 102 drivers who drove both gasoline-powered and diesel trucks for 35 years or more, the OR was 1.34 (95% CI, 0.81-2.20). There was no increased risk associated with driving both gasoline-powered and diesel vehicles for less than 35 years. Some possible limitations in this study include the validity of smoking information obtained from the next of kin, and a low response rate for questions on employment history (68%). The study results show a statistical association between >35 years of diesel exposure and lung cancer risk.
Hayes et al. (51) pooled data from three case-control studies in the United States. More than 1400 lung cancer case-control pairs were interviewed directly. Although detailed occupational data were not available, the risk of lung cancer among truck drivers employed for 10 or more years was 1.5 (95% CI, 1.1-1.9) after adjusting for daily cigarette smoking. The risk associated with other motor-vehiclerelated occupations was 1.4 (95% CI, 1.1-2.0). There was no information on the types of engines in this latter group.
Burns and Swanson (52) and Swanson et al. (53) conducted a population-based study of 3992 males with lung cancer. Cases were diagnosed between 1984 and 1987 and were identified through the Detroit cancer registry. Patients or surrogates were interviewed. Information was collected on smoking habits and occupa-tion. Over 90% of subjects who were approached responded, although only 44% of the case interviews were completed by the subjects themselves, compared to 70% of control interviews. For white, male subjects employed as drivers of heavy trucks, the smoking-adjusted risk estimate was 1.4 (95% CI, 0.8-2.4) for 1-9 years, 1.6 (95% CI, 0.8-3.5) for 10-19 years, and 2.5 (95% CI, 1.4-4.4) for 20+ years. For drivers of light trucks, the odds ratio was 1.7 (95% CI, 0.9-3.3) for 1-9 years and 2.1 (95% CI, 0.9-4.6) for 10+ years. A significant linear trend was found with increasing years of driving heavy and light trucks. No information was available on engine type. There was no increased risk for industrial equipment operators.

Dock Workers
In a study of Swedish dock workers covering the years 1950-1974, 50 lung cancer cases and 154 matched controls were compared (54). Company records on annual fuel consumption and annual machine hours were used to calculate indirect measures of diesel exposure. The response rate was 67% and many interviews were conducted among next-of-kin. Some ex-smokers were combined with nonsmokers in the analysis. The odds ratios associated with diesel exhaust among reported nonsmokers was 1.6 for medium exposure and 2.8 for high exposure. Among smokers, the odds ratio associated with diesel exhaust was 10.7 for medium exposure and 28.9 for high exposure. However, the odds ratios for smokers and nonsmokers at each level of diesel exposure had overlapping confidence intervals.

Unmeasured Confounders
The effects of several possible confounders have not been assessed in the studies discussed.
Smoking. Because cigarette smoking is the predominant cause of lung cancer, studies of diesel exhaust and lung cancer require precise statistical adjustment for cigarette smoking. This is especially important in studies showing weak associations. However, the statistical measures of smoking do not reflect with precision the actual exposure of the respiratory tract to cigarette carcinogens. A traditional measure of smoking history is pack-years, which is the product of the duration (years) and intensity of smoking (average number of cigarettes per day). Only some studies used pack-years as a statistical covariate. This measure is only an approximate method of estimating exposure to cigarette tar and particulates. Information obtained from next-of-kin adds further uncertainty in accurately classifying smoking habits. A more refined measure of smoking is total lifetime tar intake. Zang and Wynder (55) calculated that men who smoked >20 packyears have an odds ratio of lung cancer that varies from 26.9 to 48.4 depending on their lifetime tar intake.
Even lifetime tar indices are an inexact measure of cigarette carcinogen intake. The tar values are determined by the Federal Trade Commission (56) using outdated methods. The tar values for different cigarette brands are determined by smoking machines under standard laboratory conditions taking one puff per minute of 2 sec duration and a 35-ml volume. These standard conditions were established in 1936 to reflect the smoking habits of nonfiltered cigarettes. Today, most smokers use filtered cigarettes. The inhalation patterns of cigarettes differ between smokers of nonfiltered cigarettes and smokers of filtered cigarettes.
Apparently, the low relatively elevated odds ratios in studies of diesel engine exhaust and lung cancer may be confounded by incomplete statistical adjustment for smoking.
Asbestos. Truck drivers may be exposed to airborne asbestos in the driver's cab. After a clinical report of asbestosis in a truck driver (57), dust samples were taken from the cabs of 10 trucks (58). Three cabs contained airborne asbestos fibers, and seven cabs contained synthetic fibrous minerals. The airborne concentration of fibers was not determined. These fibers likely originated from the insulation materials in the cab. The joint effect of asbestos and cigarette smoking on the risk of lung cancer in truckers (59) needs to be considered.
Dietary Fat. Saturated fat is a lung tumor promoter in animal models (60)(61)(62) and has been related to the development of lung cancer in epidemiologic studies (63)(64)(65)(66)(67). International comparisons of lung cancer rates also implicate dietary fat as a lung tumor promoter. Although the prevalence of smoking has been higher in Japan than in the United States since 1955, lung cancer rates in Japan have been substantially lower (68). This paradox may reflect the lower per capita intake of dietary fat in Japan.
There is little information on the dietary habits of diesel-exposed workers. In one survey of 206 long-distance truck drivers, Wynder and Miller (69) found a high consumption of dairy products and fatty foods. Sixty percent of truckers reported eating two or more eggs per day. Their consumption of butter, margarine, and cheese was also higher than that reported in national surveys, reflecting frequent meal consumption at roadside restaurants. Other studies suggest that blue-collar workers such as skilled technicians or laborers eat more meat and fewer vegetables compared to professionals and other white-collar workers (70). Body weight. Leanness is an independent risk factor for adenocarcinoma of the lung in epidemiologic studies (71,72). The mechanisms for these observations are unknown but could reflect increased metabolic rate. The possible confounding effects of diet and body weight have not been accounted for in epidemiologic studies of diesel exhaust and lung cancer.

Conclusion
Determining whether diesel engine exhaust is a human lung carcinogen is clearly a complex undertaking (73). The results from earlier experimental, mutagenesis, and epidemiologic studies may be less relevant than more recent studies, as changes in emission control technology have altered the chemical composition of diesel exhaust. Diesel particulate extracts are mutagenic in some bioassays, although urinary mutagenic activity was not associated with levels of diesel exhaust exposure in a population of railroad workers (74). Toxicologic data suggest that short-term metabolic "overload" from diesel exposure induces lung tumors in rats. It is unknown whether this is an appropriate model for human exposure, although cigarette smoke reduces pulmonary clearance in causing human lung cancer. If the results from rat studies are predictive of human health effects, it is unclear whether cumulative, long-term exposure to diesel emissions can overload defensive pulmonary systems. The epidemiologic studies do not provide evidence for a short-term (<20 years) carcinogenic effect of diesel exhaust.
The results from epidemiologic studies are not consistent in different study populations, although some studies were done with greater precision than others. Few studies have shown a dose-dependent relationship independent of cigarette smoking. The lack of a dose effect with cumulative diesel exposure indices must also be interpreted with caution. In general, epidemiology is too imprecise a science to detect trends in weak associations.
There is statistical evidence that longterm employment (>20 years) for locomotive engineers, diesel mechanics, trainmen, and other railroad workers is associated with an increased risk of lung cancer (44). Two other studies in the railroad industry (43,48) found statistical evidence linking long-term diesel exposure to a small increase in lung cancer rates, although the lack of adjustment for smoking habits requires a cautious interpretation. There is little evidence of a dose-response relationship because only men who had the longest duration of exposure (>20 years) in the two Garshick studies had increased risks of lung cancer. The study by Howe et al. (43) had no data on duration of employment. There was no association between employment in the railroad industry and lung cancer risk in the American Cancer Society study, although only 14 deaths occurred in this group (40). Case-control studies by Burns et al. (52) and Hall and Wynder (37) found no association between railroad employment and the risk of lung cancer, although these studies lacked information on diesel exposure. In summary, there is statistical information linking long-term exposure (>20 years) to a small increased risk for locomotive engineers, brakemen, and diesel mechanics.
A limitation in the studies of truck drivers is the inadequate characterization and statistical control for cigarette smoking. This does not necessarily imply a deficiency in the data collection instruments. It is important to recognize that when cigarette smoking is a strong confounder in studies of weak associations, traditional measures of cigarette smoking may not be precise enough to allow complete control for confounding in statistical models. The various studies of truck drivers found no increased lung cancer risk with short-term employment. Steenland et al. (50) observed a statistical increase of lung cancer among diesel truck drivers who were employed for 35 or more years. In contrast, Boffetta et al. (40) found no significant increased risk with employment as (primarily) diesel truck driving in the American Cancer Society cohort study. Boffetta et al. (39) also found no increased risk of lung cancer in truckers who reported exposure to diesel exhaust in a case-control study. Hall and Wynder (3) found no increased risk for truck drivers in a separate case-control study, although the types of trucks were unspecified. Burns et al. (52) found a significant increased risk for drivers in a case-control study, although no information was available on diesel exposure.
Swanson et al. (53) found a trend with years of employment for drivers of heavy trucks, but also found a trend with drivers of light trucks. Similarly, Hayes et al. (51) found a significant increased risk for truck drivers in a case-control study, but also found a significant risk for other motorvehicle occupations besides truck driving. Doll has pointed out that internal study inconsistencies are more likely to reflect chance findings than identification of occupational risk factors (74). In this case, there may be other factors associated with driving trucks besides engine type related to lung cancer. Benhamou et al. (35) reported a significant increase for motor vehicle drivers but did not specify the type of vehicle. In summary, the findings for motor-vehicle drivers are inconsistent. Among those studies with positive findings, it is unclear whether the associations reflect an effect of diesel exposure. There is little evidence of a dose-response trend in these data. The study of Swedish dock workers is inconclusive. It was not possible to separate the independent effects of cigarette. smoking from diesel exposure.
Elemental carbon has been used as a marker of exposure to diesel exhaust in industrial hygiene surveys. The average concentration of elemental carbon in the cabins of diesel and gasoline trucks is higher than in residential environments, but not elevated above highway background levels (75,76). Because drivers of diesel trucks and drivers of gasoline trucks are exposed to similar concentrations of diesel, this could explain in part the similar risk estimates of lung cancer for these two groups in some epidemiologic studies. Average respirable concentrations of particulates are 1-7 times higher for railroad workers (17-134 pg/m3) (72) than average truck driver exposures (20 pg/m3) (75), after adjustment for smoking. These differences suggest that the low odds ratios in studies of railroad studies cannot be generalized to other diesel-exposed occupational groups. These hygiene measurements do not necessarily reflect historic levels in these industries, although the much higher exposure of railroad workers to diesel raises questions concerning the validity of the elevated risks associated with trucking. The thirty-fifth ASCB Annual Meeting will include symposia, mini symposia, poster sessions, special interest subgroup meetings, special lectures, workshops, and other events that reflect the eclectic nature of cell biology and the tremendous impact of cell biology on all aspects of biomedical research. Each facet of the program incorporates venues designed to increase interaction among scientists and the exchange of ideas among all participants.

EXHIBITS
The commercial exhibits will be open 9:00AM-4:00PM Sunday-Tuesday, December 10-12 and Wednesday, December 13 from 9:0OAM-3:00PM. There will be approximately 450 exhibit booths, allowing registrants the opportunity to examine state-of-the-art products and services. The ASCB will provide complimentary refreshments each morning and afternoon in the exhibit hall.