Workplace Diesel Exhausts and Gasoline Exposure and Risk of Colorectal Cancer in Four Nordic Countries

Original Article Workplace Diesel Exhausts and Gasoline Exposure and Risk of Colorectal Cancer in Four Nordic Countries Madar Talibov 1 , 2 , * , Jorma Sormunen 1 , 3 , Elisabete Weiderpass 4 , 5 , 6 , 7 , Kristina Kjaerheim 8 , Jan-Ivar Martinsen 8 , Per Sparen 6 , Laufey Tryggvadottir 9 , 10 , Johnni Hansen 11 , Eero Pukkala 1 , 12 1 Faculty of Social/Health Sciences, University of Tampere, Tampere, Finland 2 International Agency for Research on Cancer, Environment and Radiation Section, Lyon, France 3 Tampere University Hospital, Department of Oncology, Tampere, Finland 4 Department of Community Medicine, Faculty of Health Sciences, University of Tromsø, The Arctic University of Norway, Tromsø, Norway 5 Department of Research, Cancer Registry of Norway, Institute of Population-Based Cancer Research, Oslo, Norway 6 Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden 7 Genetic Epidemiology Group, Folkhälsan Research Center, Faculty of Medicine, University of Helsinki, Helsinki, Finland 8 Cancer Registry of Norway, Oslo, Norway 9 Icelandic Cancer Registry, Reykjavik, Iceland 10 Faculty of Medicine, University of Iceland, Reykjavik, Iceland 11 The Danish Cancer Society Research Center, Copenhagen, Denmark 12 Finnish Cancer Registry, Institute for Statistical and Epidemiological Cancer Research, Helsinki, Finland article info Article history: Received 1 June 2018 Received in revised form 14 September 2018 Accepted 2 January 2019 Available online 9 January 2019 Keywords: Caseecontrol study Colorectal cancer Diesel exhaust Gasoline Workplace abstract Background: Evidence on associations between occupational diesel exhaust and gasoline exposure and colorectal cancer is limited. We aimed to assess the effect of workplace exposure to diesel exhaust and gasoline on the risk of colorectal cancer. Methods: This caseecontrol study included 181,709 colon cancer and 109,227 rectal cancer cases diagnosed between 1961 and 2005 in Finland, Iceland, Norway, and Sweden. Cases and controls were identified from the Nordic Occupational Cancer Study cohort and matched for country, birth year, and sex. Diesel exhaust and gasoline exposure values were assigned by country-specific job-exposure matrices. Odds ratios and 95% confidence intervals were calculated by using conditional logistic regression models. The results were adjusted for physical strain at work and occupational exposure to benzene, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, and wood dust. Results: Diesel exhaust exposure was associated with a small increase in the risk of rectal cancer (odds ratio ¼1.05, 95% confidence interval 1.02-1.08). Gasoline exposure was not associated with colorectal cancer risk. Conclusion: This study showed a small risk increase for rectal cancer after workplace diesel exhaust exposure. However, this finding could be due to chance, given the limitations of the study. Ó2019 Occupational Safety and Health Research Institute, Published by Elsevier Korea LLC. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Colorectal cancer is the third most common cancer in men and the second in women with about two-thirds of the cases occurring in countries with a high human development index [1]. The incidence rate of colorectal cancer varies widely for both sexes worldwide with the highest rates observed in Australia/New Zealand and the lowest in Western Africa [1]. The incidence rate of colorectal cancer has increased in the Nordic countries over the past decades [2]. Obesity, lack of physical activity, smoking, alcohol intake, and consumption of red and processed meat are among lifestyle *Corresponding author. Faculty of Social/Health Sciences, University of Tampere, Tampere, Finland. E-mail addresses: Mader.T[email protected] (M. Talibov), [email protected] (J. Sormunen), [email protected] (E. Weiderpass), [email protected] (K. Kjaerheim), Jan.Ivar[email protected] (J.-I. Martinsen), par[email protected] (P. Sparen), [email protected] (L. Tryggvadottir), [email protected] (J. Hansen), eero.pukkala@cancer.fi(E. Pukkala). Contents lists available at ScienceDirect Safety and Health at Work journal homepage: www.e-shaw.org 2093-7911/$ esee front matter Ó2019 Occupational Safety and Health Research Institute, Published by Elsevier Korea LLC. This is an open access article under the CC BY-NCND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). https://doi.org/10.1016/j.shaw.2019.01.001 Safety and Health at Work 10 (2019) 141e150 factors that have been linked to an increased risk of colorectal cancers [3e8]. Previous studies suggested associations also with occupational agents. For example, physically active work was associated with reduced risk of colorectal cancer, particularly with distal sites of the colon [9,10]. Prolonged exposure to asbestos was linked to an elevated risk of cancer of the total colon, distal colon, and rectum in the Prospective Netherlands Study [11]. Increased risk of colorectal cancer was also linked to night shift work [12], benzene exposure [13,14], and exposure to metalworking fluids [15,16]. Evidence on associations between workplace diesel exhaust and gasoline exposure and colorectal cancer is scarce. Few studies observed modest association between workplace diesel exhaust exposure and risk of the colon and rectum [13,17e19]. The aim of the present study was to assess associations between workplace diesel exhaust and gasoline exposures and colorectal cancer, including its subtypes. 2. Materials and methods This caseecontrol study was nested within the Nordic Occupational Cancer Study (NOCCA) cohort. The NOCCA cohort includes 15 million persons from Finland, Iceland, Norway, Sweden, and Denmark, who were aged from 30 to 64 years on January 1 of the year after the first available census where they participated [20]. They were followed up until the date of emigration, death, or 31 December of the following years: 2003 in Denmark and Norway, 2004 in Iceland, and 2005 in Finland and Sweden [20]. Information on dates of death and emigration was obtained from Central Population Registers of these countries. Data from various registries were linked by using unique personal identification numbers. This method ensured a complete ascertainment of relevant events for each person included in the cohort because the possibility of error in identifiers is extremely small [20]. Data from Denmark were not included in the present study because we did not have access to individual-level records from this country. All incident colorectal cancer cases diagnosed between 1961 and 2005 in Finland, Iceland, Norway, and Sweden, and reported to the national cancer registries, were included in this study. Categories of the ascending, transversal, and descending colon were used for specific analysis. The remaining sites (e.g. sigmoid colon, appendix, cecum, splenic and hepatic flexures) were combined into the category of “other colon”. Five controls for each case were randomly selected from the NOCCA cohort. Cases and controls could have a previous history of cancer other than colorectal cancer before the date of diagnosis of the case (“index date”). Cases and controls were matched by country, sex, and the year of birth, and the controls were living in the country on the index date. Study participants had to be 20 years or older on the index date and had to have at least one census record before that date. Job titles of study participants were available from computerized census records from 1960, 1970, 1980, and 1990 in Sweden; from 1960, 1970, and 1980 in Norway; and from 1970, 1980, and 1990 in Finland. In Iceland, the only computerized census record was available from 1981 census [20]. Diesel exhaust and gasoline exposure values were assigned by linking the NOCCA job-exposure matrix (NOCCA-JEM) to job titles of study participants. The NOCCA-JEM was developed by a Nordic expert panel including experts from each country, based on the template of the Finnish job-exposure matrix [21]. It assigns prevalence of exposure (P) and annual average level (L) of exposure among the exposed persons for 28 occupational agents in more than 300 specific occupational groups in four time periods: 1945e 1959, 1960e1974, 1975e1984, and 1985e1994 [22]. We assigned a product of P and L of diesel exhaust and gasoline exposures to each year over the duration of the employment period of study participants. These values were then summed up to estimate cumulative exposures. Occupational groups exposed to diesel engine exhaust and gasoline are presented in Appendix A. The employment period of study participants was assumed to start at age 20 years (typical age at job start) and end either at age 65 years (typical retirement age) or on the index date, whichever occurred first. If a person had different occupations in different censuses, we assumed that he/ she changed occupation midway of known census years. The same procedure was used to estimate cumulative exposures for other occupational agents. Selection of covariates for the main effect model was based on the “purposeful covariate selection”method [23]. Covariates with Wald test p-value less than 0.25 from univariate logistic regression models were selected as candidates for the multivariate model. In the next step, covariates were removed from the multivariate model if they were not significantly contributing to the model fit. This procedure suggested that benzene, formaldehyde, ionizing radiation, chromium, chlorinated hydrocarbons, wood dust, and perceived physical workload could be included into the final main effect model as covariates. Because none of these covariates were strongly correlated with diesel exhaust or gasoline, we included them in the same model. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by using conditional logistic regression models. In multilevel exposure analysis, cumulative diesel exhaust and gasoline exposures were categorized by using 50 th and 90 th percentiles of exposure distribution among exposed controls as cutpoints. Hence, the resulting exposure categories were the following: unexposed, <50 th percentile, 50 th e90 th , and >90 th percentile. Unexposed categories were used as a reference in all analyses. Ordinal levels of exposure categories were used as continuous to test for significance of doseeresponse relationship. In overall exposure analysis, unexposed category was defined as never exposed, and all other categories were combined into the everexposed category. Significance of interaction among diesel exhaust, gasoline, and sex were assessed by using analysis of variance. The lifestyle-related factors by occupation and gender were available from the Finnish job-exposure matrix [21]. These data were based on the Finnish Health Behaviour and Health Among the Finnish Population surveys conducted by the Finnish Institute for Health and Welfare since 1978. The purpose of these surveys was to collect information on the health of employment-aged persons to track trends and changes over time. The main topics included in the surveys were eating habits, tobacco use, physical activity, health conditions, and alcohol consumption [24]. We controlled for the following lifestyle factors in the sensitivity analysis including only the Finnish data: the proportion of daily smokers; proportion of men drinking at least eight and women drinking five portions of alcohol weekly; proportion of those who fulfill fewer than three of the four recommended dietary habits; proportion of those who have leisure time exercise less than twice a week; and proportion of those with a body mass index of 25 or higher [21]. Other sensitivity analyses included analyses with 10and 20year lag time and analysis with tertile categorization. The lag time analyses were performed under the assumption that recent exposures may not be related to cancer risk. In 10and 20-year lag time analyses, we did not count exposures occurring 10 and 20 years before the index date, respectively. All analyses were conducted by using R statistical software, version 3.4.1 (R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/). Saf Health Work 2019;10:141e150142 3. Results Table 1 shows numbers and proportions of cases for each cancer site by country, sex, and age at diagnosis. All proportions shown for cases in this table are the same also for controls because they were matched for these characteristics. The study included 181,709 colon cancer cases and 109,227 rectal cancer cases. There were more female than male colon cancer cases, whereas rectum cancer was more common in men than in women. Most of the cases were from Sweden, whereas only less than 1% were from Iceland. The median age at diagnosis was 71 years for colon cancer and 70 years for rectal cancer (Table 1). Statistically significantly increased risk of rectal cancer (OR ¼1.05, 95% CI 1.02-1.08) and decreased risk of cancer of the descending colon (OR ¼0.88, 95% CI 0.80-0.97) were observed for overall diesel exhaust exposure (Table 2). Increased risk with the borderline significance was observed for the transversal colon (OR ¼1.05, 95% CI 1.00-1.11) and all colorectal (OR ¼1.02, 95% CI 1.00-1.03). Overall gasoline exposure did not seem to be associated with colorectal cancer risk (Table 2). Analysis with categorical exposures showed similar risk pattern as in overall analysis (Table 3). However, the only significantly increased risk for rectal cancer was observed for the medium diesel exhaust category (OR ¼1.07, 95% CI 1.04-1.11), whereas for other exposure categories, risk estimates were not significant (Table 3). Analysis with adjustment for lifestyle factors in the Finnish data showed associations of diesel exhaust exposure with rectal and all colorectal cancers with significant doseeresponse relationship Table 1 Selected demographic characteristics of incident colorectal cancer cases in the Nordic Occupational Cancer Study cohort during 1961-2005 Characteristics Ascending (n¼63,867), n (%) Transversal (n¼26,681), n (%) Descending (n¼8,863), n (%) “Other colon” a (n¼82,298), n (%) All colon (n¼1,81709), n (%) Rectum (n¼1,09227), n (%) Colorectal (n¼2,90936), n (%) Country Finland 11,003 (17.2) 4,542 (17.0) 1,452 (16.4) 11,601 (14.1) 28,598 (15.7) 19,903 (18.2) 48,501 (16.7) Iceland 185 (0.3) 77 (0.3) 94 (1.1) 760 (0.9) 1,116 (0.6) 424 (0.4) 1,540 (0.5) Norway 18,683 (29.3) 8,493 (31.8) 2,782 (31.4) 20,633 (25.1) 50,591 (27.8) 28,123 (25.7) 78,714 (27.1) Sweden 33,996 (53.2) 13,569 (50.9) 4,535 (51.2) 49,304 (59.9) 101,404 (55.8) 60,777 (55.6) 162,181 (55.7) Sex Men 27,721 (43.4) 12,556 (47.1) 4,358 (49.2) 40,402 (49.1) 85,037 (46.8) 60,019 (54.9) 145,056 (49.9) Women 36,146 (56.6) 14,125 (52.9) 4,505 (50.8) 41,896 (50.9) 96,672 (53.2) 49,208 (45.1) 145,880 (50.1) Age at diagnosis 40 997 (1.6) 381 (1.4) 181 (2.0) 1,198 (1.5) 2,757 (1.5) 984 (0.9) 3,741 (1.3) 41e60 9,947 (15.6) 5,044 (18.9) 2,045 (23.1) 16,342 (19.9) 33,378 (18.4) 21,713 (19.9) 55,091 (18.9) 61e80 39,254 (61.5) 16,466 (61.7) 5,413 (61.1) 51,028 (62.0) 112,161 (61.7) 68,909 (63.1) 181,070 (62.2) 80 13,669 (21.4) 4,790 (18.0) 1,224 (13.8) 13,730 (16.7) 33,413 (18.4) 17,621 (16.1) 51,034 (17.5) Mean, median 71, 73 70, 71 68, 69 69, 71 70, 71 69, 70 70, 71 a “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. Table 2 Odds ratios (ORs) and 95% confidence intervals (95% CIs) for ever vs never occupational diesel and gasoline exposures and colorectal cancer Exposure agent Diesel Gasoline Cancer location Case, nControl, nOR a 95% CI Case, nControl, nOR a 95% CI Ascending colon Never 57,809 290,146 61,973 310,521 Ever 6,058 29,189 1.00 0.97e1.04 1,894 8,814 1.06 0.95e1.18 Transversal colon Never 23,887 120,187 25,844 129,391 Ever 2,794 13,218 1.05 1.00e1.11 837 4,014 0.97 0.82e1.16 Descending colon Never 7,981 39,673 8,593 42,927 Ever 882 4,642 0.88 0.80e0.97 270 1,388 1.00 0.75e1.33 “Other colon” b Never 73,666 369,035 796,81 398,692 Ever 8,632 42,455 0.98 0.95e1.01 2,617 12,798 0.98 0.89e1.08 All colon Never 163,343 819,041 176,091 881,531 Ever 18,366 89,504 0.99 0.97e1.02 5,618 27,014 1.01 0.94e1.08 Rectum Never 96,574 485,074 105,546 527,915 Ever 12,653 61,061 1.05 1.02e1.08 3,681 18,220 0.93 0.86e1.02 All colorectal Never 259917 1,304,115 281,637 1,409,446 Ever 31,019 150,565 1.02 1.00e1.03 9,299 45,234 0.98 0.93e1.03 a OR estimates were adjusted for benzene, perceived physical workload, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, and wood dust. b “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. M. Talibov et al / Diesel and Gasoline Exposure and Colorectal Cancer. 143 (p<0.01) (Table 4). Notably, most of the risk estimates observed in Table 3 slightly increased away from null after adjusting for lifestyle factors. Risk of rectal cancer for categorical diesel exhaust exposure remained increased also in analyses with 10and 20-year lag time (Appendices B, C). When diesel exhaust exposure was categorized using tertile cut-off points, medium and high diesel exhaust exposure levels were significantly associated with an increased rectal cancer risk with doseeresponse relationship (OR ¼1.08, 95% CI 1.04-1.12 and OR ¼1.07, 95% CI 1.02-1.11 respectively) (Appendix D). 4. Discussion The present study showed small positive association between workplace diesel exhaust exposure and rectal cancer. This association remained increased also when adjusted for lifestyle factors, when diesel exhaust exposure was categorized using tertile cut-off points, and in analysis with 10and 20-year lag time. Observed statistically significantly decreased overall risk of descending colon cancer is likely to be a chance finding due to multiple testing as it was not confirmed in other analyses. We did not observe association between occupational gasoline exposure and colorectal cancer. Diesel and gasoline are most widely used fuel types in combustion engines, and their emissions consist of many carcinogens, including polycyclic aromatic hydrocarbons, nitroarenes, carbon monoxide, and 3-nitrobenzathrone among others [25,26]. Although similar particles are emitted from both gasolineand diesel-powered engines, the distribution and surface properties of the particles are different, suggesting potential differences in health effects associated with these exposures [27]. The main route of diesel exhaust and gasoline exposure was inhalation of polluted ambient air. Some of inhaled particles accumulated in the respiratory tract could be translocated to gastrointestinal tract as a result of mucociliar clearance [28]. Previous animal studies showed that diesel exhaust particles administered via the gastrointestinal route Table 3 Odds ratios (ORs) and 95% confidence intervals (95% CIs) for occupational diesel and gasoline exposures and colorectal cancer. Exposure agent Diesel a Gasoline b Cancer location Case, nControl, nOR c 95% CI p-trend OR c 95% CI OR c 95% CI p-trend Ascending colon Unexposed 57,809 290,146 1.00 61,973 310,521 1.00 Low 3,039 14,600 1.02 0.97e1.06 903 4,278 1.07 0.95e1.21 Medium 2,377 11,687 0.98 0.93e1.03 779 3,610 1.03 0.90e1.19 High 642 2,902 1.04 0.94e1.15 0.92 212 926 1.06 0.88e1.27 0.50 Transversal colon Unexposed 23,887 120,187 1.00 25,844 129,391 1.00 Low 1,390 6,637 1.05 0.98e1.12 414 2,050 0.95 0.79e1.16 Medium 1,124 5,232 1.07 0.99e1.15 337 1,583 0.99 0.80e1.22 High 280 1,349 1.03 0.89e1.19 0.11 86 381 1.02 0.77e1.36 0.87 Descending colon Unexposed 7,981 39,673 1.00 8,593 42,927 1.00 Low 403 2,366 0.80 0.71e0.91 137 664 1.18 0.86e1.62 Medium 388 1,819 0.97 0.86e1.11 111 575 0.90 0.63e1.29 High 91 457 0.96 0.74e1.24 0.24 22 149 0.68 0.40e1.15 0.09 “Other colon” d Unexposed 73,666 369,035 1.00 79,681 398,692 1.00 Low 4,385 21,215 1.00 0.97e1.04 1,269 6,431 0.95 0.86e1.06 Medium 3,364 17,025 0.95 0.91e0.99 1,039 5,102 0.96 0.85e1.08 High 883 4,215 1.02 0.94e1.11 0.09 309 1,265 1.15 0.98e1.34 0.29 All colon Unexposed 163,343 819,041 1.00 176,091 881,531 1.00 Low 9,217 44,818 1.00 0.98e1.03 2,723 13,423 1.00 0.93e1.08 Medium 7,253 35,763 0.98 0.95e1.01 2,266 10,870 0.98 0.91e1.07 High 1,896 8,923 1.02 0.97e1.09 0.46 629 2,721 1.07 0.97e1.19 0.43 Rectum Unexposed 96,574 485,074 1.00 105,546 527,915 1.00 Low 6,190 30,553 1.03 0.99e1.06 1,819 9,337 0.93 0.85e1.02 Medium 5,187 24,422 1.07 1.04e1.11 1,512 7,143 0.94 0.85e1.04 High 1,276 6,086 1.05 0.98e1.13 <0.01 350 1,740 0.92 0.80e1.06 0.18 All colorectal Unexposed 259,917 1,304,115 1.00 281,637 1,409,446 1.00 Low 15,410 75,382 1.02 0.99e1.04 4,535 22,732 0.97 0.92e1.03 Medium 12,441 60,195 1.02 0.99e1.04 3,784 18,028 0.97 0.91e1.04 High 3,168 14,988 1.04 0.99e1.09 0.05 980 4,474 1.02 0.94e1.11 0.94 Diesel and gasoline were categorized based on 50 th and 90 th percentile of cumulative exposure distribution among exposed colorectal cancer cases and controls. a The low diesel exposure category was defined as 0.8 milligram per cubic meter (mg/m 3 ); medium category 0.8e2.3 mg/m 3 ; and high category >2.3 mg/m 3 . The unexposed category was used as a reference. b The low gasoline exposure category was defined as 1.9 parts per million (ppm)-years; medium category 1.9e4.6 ppm-years; and high category >4.6 ppm-years. The unexposed category was used as a reference. c OR estimates were adjusted for benzene, perceived physical workload, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, and wood dust. d “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. Saf Health Work 2019;10:141e150144 can induce DNA adducts and oxidative stress resulting in DNA strand breaks in gastrointestinal epithelial cells [29,30]. Accuracy and completeness of cancer incidence data is one of the strengths of the present study. Validation studies showed high degree of completeness, comparability, accuracy, and timeliness of cancer registration in the Nordic countries [31]. Reliable occupational data from census records are another advantage of the study. Previous studies demonstrated high accuracy of occupational classifications based on census records in the Nordic countries [32,33]. Finally, by linking job histories to NOCCA-JEM, we were able to control for the effect of many concomitant agents that can be present among diesel exhaust and gasoline-exposed workers. Potential exposure misclassification is the main limitation of the study. First, the NOCCA-JEM cannot account for exposure heterogeneity within the occupation because it assigns average exposure to all members of the occupational group. Second, the NOCCA-JEM does not separate occupations by industry. Exposure intensity and prevalence may vary by industries included into the same occupation. Third, we did not have complete job histories of study participants and therefore imputed them from available computerized census records by assuming that a person changed occupation midway between consecutive censuses. Job histories were imputed from four census records in Sweden and from three census records in Finland and Norway. In Iceland, the only available computerized census record was 1981 census. However, this is unlikely to strongly bias the main results because Icelandic population constituted only less than 1% of the overall study population (Table 1). In addition, previous studies demonstrated low occupational mobility in the Nordic countries [20,32]. We could not control for leisure time physical activity, diet, smoking, alcohol intake, and body mass index in the main analyses. These factors have been linked to colorectal cancer risk in previous studies [3,4,6e8]. However, we were able to assess on the aggregate level the effect of lifestyle factors on associations between diesel Table 4 Odds ratios (ORs) and 95% confidence intervals (95% CIs) for occupational diesel and gasoline exposures and colorectal cancer in Finland. Exposure agent Diesel a Gasoline b Cancer location Case, nControl, nOR c 95% CI p-trend Case, nControl, nOR c 95% CI p-trend Ascending colon Unexposed 10,191 51,332 1.00 10,833 54,278 1.00 Low 293 1,359 1.06 0.93e1.22 78 375 1.09 0.77e1.57 Medium 418 1,894 1.05 0.93ee1.19 77 288 1.10 0.69e1.75 High 101 430 1.08 0.85e1.38 0.23 15 74 0.86 0.44e1.66 0.96 Transversal colon Unexposed 4,182 21,030 1.00 4,469 22,375 1.00 Low 115 619 0.94 0.75e1.16 34 180 0.88 0.51e1.53 Medium 195 828 1.14 0.95e1.38 32 127 0.94 0.47e1.89 High 50 233 1.04 0.73e1.46 0.37 7 28 0.98 0.36e2.69 0.83 Descending colon Unexposed 1,335 6,727 1.00 1,432 7,150 1.00 Low 37 221 0.85 0.58e1.24 8 52 0.91 0.32e2.62 Medium 62 264 1.09 0.78e1.52 11 43 0.55 0.15e2.02 High 18 48 1.79 0.97e3.32 0.22 1 15 0.16 0.01e1.72 0.16 “Other colon” d Unexposed 10,678 53,783 1.00 11,439 57,191 1.00 Low 320 1,408 1.09 0.95e1.24 72 352 1.06 0.74e1.51 Medium 470 2,212 1.01 0.90e1.14 69 371 0.85 0.54e1.32 High 133 602 1.06 0.86e1.31 0.52 21 91 0.99 0.55e1.81 0.81 All colon Unexposed 26,386 132,872 1.00 28,173 140,994 1.00 Low 765 3,607 1.04 0.95e1.13 192 959 1.03 0.83e1.30 Medium 1145 5,198 1.05 0.97e1.13 189 829 0.94 0.71e1.24 High 302 1,313 1.09 0.95e1.26 0.07 44 208 0.87 0.58e1.29 0.57 Rectum Unexposed 18,180 91,690 1.00 19,606 97,985 1.00 Low 574 2,636 1.10 1.01e1.22 145 731 0.93 0.71e1.22 Medium 918 4,155 1.11 1.02e1.21 128 638 0.83 0.60e1.15 High 231 1,034 1.14 0.97e1.34 <0.01 24 161 0.65 0.38e1.10 0.10 All colorectal Unexposed 44,566 224,562 1.00 47,779 238,979 1.00 Low 1,339 6,243 1.06 1.00e1.14 337 1,690 0.99 0.84e1.18 Medium 2,063 9,353 1.08 1.02e1.14 317 1,467 0.91 0.74e1.13 High 533 2,347 1.11 1.00e1.23 <0.01 68 369 0.77 0.56e1.06 0.17 Diesel and gasoline were categorized based on 50 th and 90 th percentile of cumulative exposure distribution among exposed colorectal cancer cases and controls. a The low diesel exposure category was defined as 0.8 milligram per cubic meter (mg/m 3 ); medium category 0.8e2.3 mg/m 3 ; and high category >2.3 mg/m 3 . The unexposed category was used as a reference. b The low gasoline exposure category was defined as 1.9 parts per million (ppm)-years; medium category 1.9e4.6 ppm-years; and high category >4.6 ppm-years. The unexposed category was used as a reference. c OR estimates were adjusted for benzene, perceived physical workload, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, wood dust, smoking, alcohol, BMI, diet, and physical activity. d “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. M. Talibov et al / Diesel and Gasoline Exposure and Colorectal Cancer. 145 exhaust, gasoline, and colorectal cancer in the Finnish part of the data. Adjustment for lifestyle factors slightly increased risk estimates away from the null. Therefore, if data on lifestyle factors were available, most of the associations observed in the main analysis would likely to be stronger. Associations observed in this study are consistent but weaker than the results from studies conducted in Canada [13,18,19], which observed increased risk of rectal cancer. A recent Australian casee control study [34] reported nonsignificantly increased risk of all colorectal cancer for exposure to diesel and gasoline exhaust emissions (OR ¼1.14, 95% CI 0.89-1.46 for diesel and OR ¼1.07, 95% CI 0.84-1.36 for gasoline). The difference in results between our study and Canadian and Australian studies could in part be explained by the difference in the prevalence of diesel exhaust exposure. For example, any exposure to diesel exhaust was only 11% in our study population compared with 19% in an Australian study [34] and 18% and 36% in Canadian studies [18,19]. In conclusion, the present study showed a small risk increase of rectal cancer among workers occupationally exposed to diesel exhaust. However, we cannot exclude the possibility of this weak to modest association to be due to chance, given the limitations of the present study. Workplace gasoline exposure was not linked to colorectal cancer risk. Ethical approval As this study was register-based, neither ethical committee review nor informed consent from the study participants was required. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Conflict of interest Authors declare no conflict of interest. Acknowledgment The authors thank the Nordic Occupational Cancer Studies (NOCCA) project members for the development of NOCCA cohort data and job-exposure matrix. Appendix Appendix A Occupational groups exposed to diesel engine exhaust and gasoline. Occupation a 1945e1959 1960e1974 1975e1984 1985e94 (P x L) b (P x L) b (P x L) b (P x L) b Diesel engine exhaust c Miners and quarrymen 0.03 0.08 0.33 0.28 Asphalt workers 0.10 0.14 0.13 0.10 Construction machinery operators 0.07 0.13 0.12 0.09 Railway engine drivers, steam engine firemen 0.02 0.10 0.09 0.05 Engine room crew 0.10 0.07 0.07 0.07 Road transport supervisors 0.07 0.09 0.08 0.07 Harbor masters 0.06 0.08 0.07 0.06 Road and tram service personnel 0.05 0.08 0.07 0.06 Machine and engine mechanics 0.05 0.07 0.07 0.06 Stevedores 0.02 0.07 0.06 0.04 Maintenance crews and supervisors 0.02 0.06 0.05 0.04 Assisting construction workers 0.04 0.06 0.04 0.02 Messengers and delivery boys 0.02 0.05 0.05 0.04 Forklift operators 0.01 0.05 0.04 0.03 Service station attendants 0.03 0.03 0.05 0.00 Motor vehicle and tram drivers 0.02 0.03 0.03 0.03 Mechanical engineers 0.01 0.01 0.01 0.02 Policemen 0.01 0.01 0.01 0.01 Stationary engine and machinery operators 0.00 0.00 0.01 0.01 Gasoline b Service station attendants 0.80 0.80 0.06 0.00 Printers 0.30 0.00 0.00 0.00 Occupation in graphics 0.20 0.00 0.00 0.00 Machine and engine mechanics 0.10 0.10 0.02 0.01 Estimates were retrieved from the Nordic Occupational Cancer Study job-exposure matrix. a Occupations were listed from the largest to the smallest P L value. b P was proportion, and L was annual average exposure in the occupational group. c The unit of diesel engine exposure was mg/m 3 , and the unit of gasoline exposure was parts per million (ppm). Saf Health Work 2019;10:141e150146 Appendix B Odds ratios (ORs) and 95% confidence intervals (95% CIs) for occupational diesel and gasoline exposures and colorectal cancer from 10-year lag time analysis. Exposure agent Diesel a Gasoline b Cancer location Case, nControl, nOR c 95% CI p-trend Case, nControl, nOR c 95% CI p-trend Ascending colon Unexposed 58,161 291,962 1.00 62,123 311,191 1.00 Low 3,524 16,768 1.02 0.98e1.07 783 3,751 1.04 0.92e1.18 Medium 1,961 9,535 0.97 0.91e1.03 843 3,847 1.04 0.91e1.19 High 221 1,070 1.01 0.87e1.17 0.95 118 546 0.93 0.74e1.17 0.91 Transversal colon Unexposed 24,027 120,986 1.00 25,892 129,667 1.00 Low 1,661 7,618 1.08 1.02e1.15 380 1,831 1.02 0.84e1.24 Medium 888 4,268 1.03 0.94e1.13 363 1,696 1.02 0.83e1.25 High 105 533 1.00 0.81e1.24 0.13 46 211 0.98 0.68e1.42 0.83 Descending colon Unexposed 8,042 40,012 1.00 8,611 43,043 1.00 Low 498 2,704 0.85 0.76e0.95 121 567 1.29 0.94e1.78 Medium 281 1451 0.87 0.74e1.03 119 618 0.97 0.69e1.36 High 42 148 1.30 0.91e1.86 0.16 12 87 0.58 0.29e1.15 0.24 “Other colon” d Unexposed 74,161 371,640 1.00 79,871 399,621 1.00 Low 5,039 24,278 1.00 0.97e1.04 1,114 5,692 0.95 0.85e1.06 Medium 2,748 14,039 0.93 0.88e0.98 1,128 5,385 1.02 0.91e1.15 High 350 1,533 1.11 0.99e1.25 0.22 185 792 1.11 0.92e1.35 0.54 All colon Unexposed 164,391 824,600 1.00 176,497 883,522 1.00 Low 10,722 51,368 1.01 0.99e1.04 2,398 11,841 1.01 0.93e1.08 Medium 5,878 29,293 0.96 0.92e0.99 2,453 11,546 1.03 0.95e1.11 High 718 3,284 1.07 0.98e1.16 0.61 361 1,636 1.00 0.88e1.14 0.76 Rectum Unexposed 97,221 488,847 1.00 105,785 529,241 1.00 Low 7,376 35,083 1.07 1.04e1.10 1,631 8,258 0.96 0.88e1.05 Medium 4,140 19,934 1.05 1.00e1.10 1,604 7,530 0.97 0.88e1.06 High 490 2,271 1.09 0.99e1.21 <0.01 207 1,106 0.84 0.71e0.99 0.13 All colorectal Unexposed 261,612 1,313,447 1.00 282,282 1,412,763 1.00 Low 18,098 86,451 1.04 1.02e1.05 4,029 20,099 0.99 0.93e1.05 Medium 10,018 49,227 0.99 0.97e1.02 4,057 19,076 1.01 0.95e1.07 High 1,208 5,555 1.08 1.01e1.15 0.02 568 2,742 0.94 0.84e1.04 0.55 Diesel and gasoline were categorized based on 50 th and 90 th percentile of cumulative exposure distribution among exposed colorectal cancer cases and controls. a The low diesel exposure category was defined as 0.8 milligram per cubic meters (mg/m 3 ); medium category 0.8e2.3 mg/m 3 ; and high category >2.3 mg/m 3 . The unexposed category was used as a reference. b The low gasoline exposure category was defined as 1.9 parts per million (ppm)-years; medium category 1.9e4.6 ppm-years; and high category >4.6 ppm-years. The unexposed category was used as a reference. c OR estimates were adjusted for benzene, perceived physical workload, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, and wood dust. d “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. M. Talibov et al / Diesel and Gasoline Exposure and Colorectal Cancer. 147 Appendix C Odds ratios (ORs) and 95% confidence intervals (95% CIs) for occupational diesel and gasoline exposures and colorectal cancer from 20-year lag time analysis. Exposure agent Diesel a Gasoline b Cancer location Case, nControl, nOR c 95% CI p-trend Case, nControl, nOR c 95% CI p-trend Ascending colon Unexposed 59,019 296,053 1.00 62,428 312,619 1.00 Low 3,535 16,969 1.01 0.96e1.05 566 2,731 1.02 0.90e1.16 Medium 1,269 6,077 0.98 0.90e1.07 776 3,534 1.03 0.90e1.17 High 44 236 0.91 0.66e1.26 0.92 97 451 0.89 0.69e1.15 0.90 Transversal colon Unexposed 24,394 122,859 1.00 26,021 130,326 1.00 Low 1,693 7,701 1.08 1.02e1.15 296 1,395 1.04 0.86e1.25 Medium 573 2,708 1.08 0.95e1.22 328 1,516 1.01 0.83e1.23 High 21 137 0.78 0.49e1.23 0.04 36 168 0.95 0.63e1.43 0.91 Descending colon Unexposed 8,155 40,661 1.00 8,661 43,274 1.00 Low 511 2,732 0.86 0.77e0.96 89 405 1.13 0.82e1.55 Medium 186 889 1.04 0.83e1.29 102 560 0.80 0.57e1.12 High 11 33 1.57 0.78e3.16 0.37 11 76 0.60 0.29e1.21 0.09 “Other colon” d Unexposed 75,296 377,383 1.00 80,297 401,663 1.00 Low 5,127 24,794 0.99 0.96e1.03 820 4,314 0.92 0.82e1.02 Medium 1,804 8,980 0.98 0.91e1.05 1023 4,846 1.00 0.90e1.12 High 71 333 1.03 0.79e1.33 0.35 158 667 1.11 0.91e1.37 0.68 All colon Unexposed 166,864 836,956 1.00 177,407 887,882 1.00 Low 10,866 52,196 1.00 0.98e1.03 1771 8,845 0.98 0.91e1.05 Medium 3,832 18,654 1.00 0.95e1.05 2,229 10,456 1.00 0.93e1.08 High 147 739 0.97 0.81e1.16 0.92 302 1,362 0.98 0.85e1.14 0.91 Rectum Unexposed 98,864 496,883 1.00 106,372 532,135 1.00 Low 7,602 35,982 1.07 1.04e1.10 1,253 6,278 0.94 0.86e1.02 Medium 2,639 12,738 1.04 0.98e1.11 1,421 6,786 0.93 0.85e1.03 High 122 532 1.17 0.96e1.42 <0.01 181 936 0.84 0.70e1.01 0.03 All colorectal Unexposed 265,728 1,333,839 1.00 283,779 1,420,017 1.00 Low 18,468 88,178 1.03 1.01e1.05 3,024 15,123 0.96 0.91e1.02 Medium 6,471 31,392 1.02 0.98e1.05 3,650 17,242 0.98 0.92e1.04 High 269 1,271 1.05 0.92e1.20 0.01 483 2,298 0.93 0.83e1.04 0.16 Diesel and gasoline were categorized based on 50 th and 90 th percentile of cumulative exposure distribution among exposed colorectal cancer cases and controls. a The low diesel exposure category was defined as 0.8 milligram per cubic meter (mg/m 3 ); medium category 0.8e2.3 mg/m 3 ; and high category >2.3 mg/m 3 . The unexposed category was used as a reference. b The low gasoline exposure category was defined as 1.9 parts per million (ppm)-years; medium category 1.9e4.6 ppm-years; and high category >4.6 ppm-years. The unexposed category was used as a reference. c OR estimates were adjusted for benzene, perceived physical workload, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, and wood dust. d “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. Saf Health Work 2019;10:141e150148 Appendix E. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.shaw.2019.01.001. References [1] Ferlay JSI, Ervik M, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F. GLOBOCAN 2012 v1.0, cancer incidence and mortality worldwide: IARC CancerBase No. 11. Internet. Lyon: France International Agency for Research on Cancer. 2013. Available from: http://globocan.iarc.fr. [2] Engholm GFJ, Christensen N, Hansen HL, Hertzum-Larsen R, Johannesen TB, Kejs AMT, Khan S, Ólafsdóttir E, Petersen T, Schmidt LKH, Virtanen A, Storm HH. Cancer incidence, mortality, prevalence and survival in the nordic countries, version 8.0 (20.12.2017): association of the nordic cancer registries. Danish Cancer Society. 2017. Available from: http://www.ancr.nu. [3] Aune D, Chan DS, Vieira AR, Navarro Rosenblatt DA, Vieira R, Greenwood DC, et al. Red and processed meat intake and risk of colorectal adenomas: a systematic review and meta-analysis of epidemiological studies. Cancer Causes Control 2013;24(4):611e27. [4] Liang PS, Chen TY, Giovannucci E. Cigarette smoking and colorectal cancer incidence and mortality: systematic review and meta-analysis. Int J Cancer 2009;124(10):2406e15. [5] Nilsen TI, Romundstad PR, Petersen H, Gunnell D, Vatten LJ. Recreational physical activity and cancer risk in subsites of the colon (the Nord-Trondelag Health Study). Canc Epidemiol Biomark Prev A Pub Am Assoc Canc Res Cospons Am Soc Prev Oncol 2008;17(1):183e8. [6] Perera PS, Thompson RL, Wiseman MJ. Recent evidence for colorectal cancer prevention through healthy food, nutrition, and physical activity: implications for recommendations. Curr Nutr Rep 2012;1(1):44e54. [7] Schlesinger S, Aleksandrova K, Abar L, Vieria AR, Vingeliene S, Polemiti E, et al. Adult weight gain and colorectal adenomas-a systematic review and metaanalysis. Ann Oncol Off J Eur Soc Med Oncol 2017;28(6):1217e29. [8] Vieira AR, Abar L, Chan DSM, Vingeliene S, Polemiti E, Stevens C, et al. Foods and beverages and colorectal cancer risk: a systematic review and metaanalysis of cohort studies, an update of the evidence of the WCRF-AICR Continuous Update Project. Ann Oncol Off J Eur Soc Med Oncol 2017;28(8): 1788e802. [9] Pahwa M, Harris MA, MacLeod J, Tjepkema M, Peters PA, Demers PA. Sedentary work and the risks of colon and rectal cancer by anatomical sub-site in the Canadian census health and environment cohort (CanCHEC). Canc Epidemiol 2017;49:144e51. Appendix D Odds ratios (ORs) and 95% confidence intervals (95% CIs) for occupational diesel and gasoline exposures and colorectal cancer. Exposure agent Diesel a Gasoline b Cancer location Case, nControl, nOR c 95% CI p-trend Case, nControl, nOR c 95% CI p-trend Ascending colon Unexposed 57,809 290,146 1.00 61,973 310,521 1.00 Low 2,089 9,741 1.05 1.00e1.10 422 2,026 1.06 0.92e1.23 Medium 1,959 9,701 0.98 0.92e1.03 789 3,728 1.09 0.96e1.24 High 2,010 9,747 0.97 0.92e1.04 0.54 683 3,060 1.02 0.89e1.18 0.40 Transversal colon Unexposed 23,887 120,187 1.00 25,844 129,391 1.00 Low 924 4,439 1.03 0.95e1.11 179 987 0.84 0.67e1.05 Medium 948 4,359 1.09 1.01e1.17 356 1,722 0.99 0.81e1.22 High 922 4,420 1.04 0.95e1.14 0.13 302 1,305 1.09 0.88e1.35 0.32 Descending colon Unexposed 7,981 39,673 1.00 8,593 42,927 1.00 Low 274 1,593 0.82 0.72e0.95 58 339 1.00 0.69e1.46 Medium 297 1,540 0.89 0.78e1.02 134 542 1.28 0.92e1.78 High 311 1,509 0.95 0.81e1.12 0.18 78 507 0.72 0.49e1.04 0.19 “Other colon” d Unexposed 73,666 369,035 1.00 79,681 398,692 1.00 Low 3,024 14,217 1.03 0.99e1.08 594 3,014 0.98 0.87e1.11 Medium 2,745 14,091 0.94 0.90e0.98 1,113 5,505 0.99 0.89ee1.11 High 2,863 14,147 0.97 0.93e1.03 0.06 910 4,279 0.98 0.87e1.10 0.92 All colon Unexposed 163,343 819,041 1.00 176,091 881,531 1.00 Low 6,311 29,990 1.03 0.99e1.06 1,253 6,366 0.99 0.91e1.07 Medium 5,949 29,691 0.97 0.94e1.00 2,392 11,497 1.04 0.96e1.12 High 6,106 29,823 0.98 0.95e1.02 0.18 1,973 9,151 0.99 0.92e1.08 0.53 Rectum Unexposed 96,574 485,074 1.00 105,546 527,915 1.00 Low 4,127 20,611 1.01 0.97e1.05 856 4,579 0.88 0.79e0.99 Medium 4,238 19,983 1.08 1.04e1.12 1,567 7,798 0.91 0.83e1.01 High 4,288 20,467 1.07 1.02e1.11 <0.01 1,258 5,843 0.98 0.89e1.09 0.58 All colorectal Unexposed 259,917 1,304,115 1.00 281,637 1,409,446 1.00 Low 10,400 50,456 1.02 0.99e1.04 2,969 15,338 0.94 0.88e1.00 Medium 10,228 49,851 1.02 0.99e1.04 3,100 14,914 1.00 0.94e1.07 High 10,391 50,258 1.01 0.99e1.04 0.14 3,230 14,982 1.00 0.94e1.07 0.63 Diesel and gasoline were categorized based on tertiles of cumulative exposure distribution among exposed colorectal cancer cases and controls. a The low diesel exposure category was defined as 0.5 milligram per cubic meter (mg/m 3 ); medium category 0.5e1.1 mg/m 3 ; and high category >1.1 mg/m 3 . The unexposed category was used as a reference. b The low gasoline exposure category was defined as 1 parts per million (ppm)-years; medium category 1e3.2 ppm-years; and high category >3.2 ppm-years. The unexposed category was used as a reference. c OR estimates were adjusted for benzene, perceived physical workload, formaldehyde, ionizing radiation, chlorinated hydrocarbons, chromium, and wood dust. d “Other colon”included sigmoid colon, appendix, cecum, and splenic and hepatic flexures. M. Talibov et al / Diesel and Gasoline Exposure and Colorectal Cancer. 149