Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
The role of active smoking in the etiology of adult leukemia is now well recognized for acute myeloid leukemia (AML), but a link with other subtypes is inconsistent.1 Tobacco smoke contains substances that are known leukemogens,1 and experimental studies have shown that tobacco has an effect on the immune system,2,3 which may lead to development of lymphoid neoplasms. Tobacco smoke affects not only people who smoke but also those who are exposed to the combustion products of other people’s tobacco (passive smokers).4 Environmental tobacco smoke (ETS) is a combination of cigarette side stream smoke and mainstream smoke. Mainstream smoke is produced when cigarette smoke is inhaled, stays a few seconds in the lungs where some of the constituents (nicotine, carbon monoxide, and particulate matter) are scrubbed from the smoke, and then exhaled. Side stream smoke is the smoke produced by the burning cigarette itself. Most of ETS is side-stream smoke (about 85%); the remaining (15%) is mainstream smoke.5
The role of ETS exposures in the causation of human cancers has been extensively evaluated during the last 2 decades. Recently, the International Agency for Research on Cancer (IARC) classified ETS as a known human lung carcinogen on the basis of the aggregate evidence from more than 50 epidemiologic studies, together with knowledge of the nature of side stream and mainstream smoke, the materials absorbed during passive smoking and the quantitative relationships between dose and effect.1 For other cancer types, such as breast, bladder, gastrointestinal, and childhood cancers, IARC considers the findings to be inconclusive.1 To date, the epidemiologic literature about ETS exposure and its relation to leukemia development is limited, and the studies conducted on this association have discussed only the effect of parental smoke on childhood leukemia. Three large childhood case-control studies from the United States,6 United Kingdom,7 and Italy8 have reported no increased risk of leukemia from childhood exposure. Other studies, however, have reported increased risk for both acute lymphoid and myeloid leukemia in children exposed to parental smoking.9–12 Given the few data available, we decided to conduct an epidemiologic study to investigate the impact of ETS exposures on the risk of adult leukemia.
We conducted a population-based case–control study of adult leukemia using data from the Canadian national enhanced cancer surveillance system (NECSS). Methodology for the NECSS has been described in detail elsewhere.13 Briefly, the NECSS was a collaborative project between Health Canada and the provincial cancer registries, designed to provide an understanding of the environmental determinants of cancer. The NECSS collected data on individual risk factors from a sample of 20,755 Canadians recently diagnosed with one of 19 types of cancer, as well as from a sample of 5039 population controls with an age and sex distribution similar to the overall age/sex structure of the cancer cases in 8 of the 10 Canadian provinces (Prince Edward Island, Nova Scotia, Manitoba, Saskatchewan, British Columbia, Alberta, Newfoundland, and Ontario) between 1994 and 1997.
Our study is based on adult leukemia cases and controls from the NECSS database. During the period 1994 through 1997, the provincial cancer registries identified 1997 recently diagnosed and histologically confirmed adult leukemia cases, age 20–74 years, in the participating 8 Canadian provinces. Because of cases’ deaths (15%), and physician refusal to give consent to contact some severely ill cases (8%), only 1545 cases were sent questionnaires. The questionnaires were sent to cases within 1 to 4 months of diagnosis. A total of 1068 cases, representing 54% of all cases ascertained and 70% of cases contacted, completed and returned the questionnaire. On the basis of the International classification of diseases for Oncology, 358 had acute leukemia (307 with AML and 51 with acute lymphocytic leukemia [ALL], 643 had chronic leukemia (169 with chronic myelocytic leukemia [CML], 410 with chronic lymphocytic leukemia [CLL], and 64 with hairy cell leukemia [HCL], and 67 had leukemia not otherwise specified [NOS]).
Population controls were identified through frequency matching with all 19 types of cancer included in the NECSS database. The procedure of control selection has been reported in detail.14 In brief, the strategies used for selecting controls varied according to data accessibility in each province. Provincial health insurance registration databases were used in British Columbia, Saskatchewan, Manitoba, Prince Edward Island, and Nova Scotia. Ontario used the Ontario Ministry of Finance Property Assessment Database. Random-digit dialing was used in Alberta and Newfoundland. In total, questionnaires were mailed to 8060 individuals selected as potential controls in the 8 provinces. For 573 of these controls (7%), the questionnaires were returned because the address was incorrect, and no updated address could be found through publicly-available sources. In all, 5039 controls returned completed questionnaires, representing 67% of those contacted and 63% of those ascertained. To maximize power, we used all 5039 controls.
Mailed questionnaires, with telephone follow-up when necessary for clarification, were used to obtain information on subjects’ residential and occupational histories and on other risk factors for cancer. The questionnaire included questions about age, sex, ethnicity, educational level, family income, height, weight, residential and job histories, active and passive smoking, alcohol use, source of drinking water, occupation, physical activity, dietary history, and occupational exposure to specific carcinogens. The NECSS began collecting information on April 1, 1994, in 7 Canadian provinces (British Columbia, Alberta, Saskatchewan, Manitoba, Prince Edward Island, Nova Scotia, and Newfoundland) and on May 1, 1995, in an 8th province (Ontario). By July 31, 1997, data collection was complete in the 8 provinces for leukemia cases and the selected population controls. To avoid intrusion on ill patients or their families, the subjects were approached only if the attending physician gave permission. The NECSS protocol was reviewed and approved by the Cancer Registries’ human subjects review board in each province.
The NECSS questionnaire collected a lifetime history of residential and occupational passive smoking exposure. For every Canadian residence where a subject (case or control) had lived for at least 1 year, the subject was asked the address, first and last year of residence, and how many regular smokers usually had lived in the subject’s home. Also, for each job held for at least 1 year, data were collected about the jobs, years employed, and how many people had smoked regularly in the subject’s immediate work area. We restricted our analysis to never-smokers, defined as those who reported smoking fewer than 100 cigarettes in their lifetime. Of the 1068 leukemia cases and 5039 controls, 376 cases (35%) and 1938 controls (38%) were classified as never smokers. To reduce the potential misclassification of passive smoke exposure status, we further restricted our study analyses to subjects who reported their residential ETS exposure history for at least 75% of their lifetime (266 cases and 1326 controls). We used the following 2 indices as measures for residential and occupational exposure: (1) lifetime duration of ETS exposure (ie, total years of passive smoking exposure) and (2) lifetime smoker-years index (number of years of exposure times the number of regular smokers). For example, one smoker-year in the residence means living in the same home with one smoker for 1 year.15,16 The measures of residential and occupational exposure were also summed to form combined measures. The subgroup of cases who reported no residential or occupational ETS exposure was the reference group. ETS exposure was categorized into tertiles according to the distribution of exposure in the controls.
The risk of leukemia is known to vary with time after exposures to benzene17 and radiation.18 We therefore examined the effect of specific temporal patterns of exposure to ETS on the risk of adult leukemia. These included residential exposures in childhood only (age 18 years or younger) and adult life (19 years and older). In addition, we examined 3 time windows of residential and occupational ETS exposure (1–10, 11–20, and 21–30 years) before diagnosis for cases and before the date of interview for controls. Within the analyses of life stages and time windows, we also excluded subjects reporting less than 75% of their ETS exposure history. Allowing for a hypothesized 5-year latency period between the development of adult leukemia and its clinical recognition, we repeated the cumulative exposure analyses while excluding the last 5 years of ETS exposure.
On the basis of leukemia literature19–21 and the data available in the NECSS questionnaire, the following covariates were included in the study analyses to explore their confounding effects: age (<30, 30–39, 40–49, 50–59, and ≥60 years), sex, ethnicity (European descent, African descent, and others), family income (3 categories), educational years (≤9, 10–15, and >15 years), residence (average proportion of time living in urban areas), body mass index (<25, 25 to <29.9, and 30+ kg/m2), and occupational exposure to benzene and ionizing radiation. The occupational history of each included subject in the NECSS database was coded to Canadian 1980 Standard Industrial Codes and Canadian 1980 Standard Occupational Codes. Using Canadian 1980 Standard Occupational Codes, we linked our study data with a database derived from occupational Montreal studies22 to construct a simple job exposure matrix for benzene and ionizing radiation exposures. Using this matrix, subjects with “no” or “possible” job exposure were combined as nonexposed, whereas those with “probable” or “certain” job exposures were combined as exposed. Subjects whose occupational codes were not addressed in Montreal studies or who had missing codes were classified on the basis of their reported job titles as exposed or not exposed by one of the authors (P.A.) and an occupational hygienist, without any knowledge of case or control status.
We compared the distribution of demographic and confounding factors between controls and each of the studied leukemia subtypes using the chi-square test for the categorical variables and t test for the continuous variables. We used multivariate logistic regression models to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association of adult leukemia and its subtypes with the studied risk factors23 while controlling for the possible covariates. Potential confounding variables considered in this study were age, sex, ethnicity, residence, educational level, body mass index, and occupational exposure to benzene and ionizing radiation. To select the factors to be included in each model, we initially used stepwise regression with P value of 0.10 as entry criterion and P value of 0.15 as exclusion criterion. Accordingly, we have adjusted the final models by only subsets of the aforementioned variables. Despite stepwise regression results, we retained in final models those variables known to be biologically important in adult leukemia development (age and occupational exposure to benzene and ionizing radiation). We also retained sex in all final models because of the substantial sex differences observed between cases and controls (Table 1). Tests for trend with respect to ETS exposure were examined by the likelihood ratio statistic, assigning an ordinal value to each level of categorical variable and treating the variable as continuous in logistic regression models.
Main Characteristics of Leukemia Cases and Controls
A total of 266 nonsmoker leukemia cases (131 men and 135 women), and 1326 nonsmoker controls (464 men and 862 women), who reported at least 75% of their lifetime residential exposure history were included in the analyses. The numbers of specific leukemia subtypes were 72 AML, 18 ALL, 42 CML, 96 CLL, and 24 HCL. Excluding 22 cases and 163 controls missing any data on occupational ETS exposure, the analysis of occupational exposure comprised 244 cases and 1163 controls. Table 1 summarizes the main characteristics of cases and controls. Control patients differed from the combined leukemia cases in distribution by sex, body mass index, family income, and occupational exposure to benzene. Compared with control subjects, the proportion of cases exposed occupationally to benzene was higher for all leukemia combined, and for AML, CML, and HCL. With the exception of AML, each leukemia subtype had higher income than controls. However, we excluded income from further analyses because it was missing for more than 25% of respondents. Moreover, some variations existed among the subtypes of leukemia by age, sex, body mass index and residence. As expected, ALL cases tended to be younger than other leukemia subtypes, and CLL cases tended to be the oldest. Women predominated among AML and ALL while the sexes were equally distributed in CML, and men comprised the majority of HCL and CLL. The highest proportion of obesity (body mass index ≥30 kg/m2) was among AML cases.
Table 2 presents the association between lifetime ETS exposure and all leukemia combined. There was a tendency towards higher risks with higher ETS, more consistently with the occupational exposure. Occupational ETS exposure above 21 years was associated with an OR of 1.6 (95% CI = 1.1–2.3).
Tables 3 and 4 show the risk for leukemia subtypes associated with lifetime residential and occupational ETS exposures. There was no clear association for most of the subtypes, in particular for AML. However, residential and occupational ETS exposures were both associated with increased risk for CLL, with a positive linear trend. The adjusted odds ratio with the highest tertile of residential exposure was 2.0(1.0–4.0) for duration of exposure and 2.3 (1.2–4.5) for smoker-years index. For occupational exposure, the adjusted OR for CLL was 2.4 (1.3–4.3) for duration of exposure and 2.4 (1.3–4.3) for smoker-years index.
When we examined the risk of CLL by ETS exposure at different life stages, we found increased risk with exposures during either childhood or adult life (data not shown). However, a dose-response was evident only with adult life exposure. Comparing the previous 1–10, 11–20, and 21–30 years, no specific ETS exposure window was associated with a particularly higher risk (table available with the electronic version of this article). Repeating the cumulative exposure analyses while excluding the last 5 years of exposure (as a hypothesized latency period for adult leukemia), we did not find any differences in the observed results, particularly for CLL (data not shown).
The risk of all leukemia combined was slightly associated with ETS exposures. Within leukemia subtypes, we found no clear association except with CLL. The risk of CLL increased with higher residential and occupational ETS exposure indices, considered either separately or combined, with a positive linear trend. The highest risk estimates for CLL were with occupational exposure indices. Higher intensity of ETS exposure in the workplace has been previously observed.24 Recent epidemiologic studies on ETS exposure and the risk of lung cancer also have reported higher risks with workplace exposure than with residential exposure.16,25 In our study, the risk of CLL associated with residential ETS increased with exposure in childhood as well as adulthood. Also, the time-window exposure analyses did not find higher risk with more recent exposures. In contrast, analyses of specific occupational carcinogens (including benzene17 and ionizing radiation18) suggest that the risk of leukemia tends to increase with more recent exposure. This difference might be attributed to different mechanisms of ETS carcinogens in induction of leukemia.
The recent IARC report on tobacco smoke and adult cancer has established the role of active smoking in the development of AML, but data are inconsistent for other leukemia subtypes.1 The lack of clear and consistent positive associations in this study for AML may be explained by the small number of cases, attributable to low participation rate. There could be selection bias if nonrespondent cases differ from those analyzed in the frequency distribution of ETS exposures. Furthermore, our analyses did not include deceased and severely ill cases. Since the aggressiveness of AML has been associated with duration of smoking,26 the risk of AML might be attenuated artificially if the same is true for ETS exposure. Although the positive findings observed for CLL are somewhat different from those reported by IARC for active smoking, some studies have found an increased risk for CLL in association with tobacco smoking.27,28
Our study was population-based, with lifetime measures of both residential and occupational ETS exposure, accurate identification of leukemia cases by the provincial cancer registries, and the ability to examine the association by histologic type according to International Classification of Diseases for Oncology. Information in the NECSS database has allowed us to control for many potential confounders. Restricting our study analyses to subjects who reported at least 75% of their lifetime ETS exposure history has likely reduced the risk of exposure misclassification.
The study was limited by the low proportion of eligible cases and controls included in the analysis. The small sample size for some subtypes restricts the precision of our risk estimates. Recall bias is always a threat in case-control studies. Although it is possible that cases have recalled their ETS exposure history in more detail than control subjects, passive smoking is not widely regarded as a risk factor for adult leukemia. Influence of recall might be minimal. In our study, we found family income levels to be associated with disease. If this is linked to a selection bias, ignoring this variable may produce misleading risk estimates. However, there was no confounding by family income when we restricted the analyses to subjects with available family income data (194 cases and 972 controls).
The possibility of misclassification of ETS exposure is another potential source of bias.29 We had no direct biochemical validation of ETS (eg, urinary cotinine). However, such exposure could validate a recent ETS exposure, which is less important than long-term exposure. Passive smoking exposure may be difficult to measure accurately, because the biologic dose a subject receives depends on a large number of factors, such as the amount of time subjects spent outdoors and house size.30 As we have no information on these factors, we would expect misclassification of biologic dose to have influenced our results. Because 15% of the eligible cases had died before they could be recruited and 8% were too ill to participate, our results may be generalizable only to less aggressive leukemia cases, or to those be diagnosed earlier or responding better to treatment. In summary, these data suggest that the risk of CLL may be increased with passive exposure to cigarette smoke. Small sample size, possible misclassification of exposure, and possible selection bias limit the interpretation of our findings. Further studies are needed to confirm this result.
We thank Yang Mao of the Public Health Agency of Canada, Jack Siemiatycki, and Marie-Élise Parent of the INRS-Institut Armand-Frappier. We also thank Michel Legris, Direction de Santé Publique de la Capitale Nationale de Québec. The Canadian Cancer Registries Epidemiology Research Group is comprised of a Principal Investigator from each of the Provincial Cancer Registries involved in the National Enhanced Cancer Surveillance System: Bertha Paulse, Newfoundland Cancer Foundation; Ron Dewar, Nova Scotia Cancer Registry; Dagny Dryer, Prince Edward Island Cancer Registry; Nancy Kreiger, Cancer Care Ontario; Erich Kliewer, Cancer Care Manitoba; Diane Robson, Saskatchewan Cancer Foundation; Shirley Fincham, Division of Epidemiology, Prevention and Screening, Alberta Cancer Board; and Nhu Le, British Columbia Cancer Agency.
1. IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Tobacco Smoking and Involuntary Smoking. International Agency for Research on Cancer. Volume 83. Lyon: IARC; 2004.
2. Geng Y, Savage SM, Razani-Boroujerdi S, et al. Effects of nicotine on the immune response. II. Chronic nicotine treatment induces T cell anergy. J Immunol
3. Geng Y, Savage SM, Johson LJ, et al. Effects of nicotine on the immune response. I. Chronic exposure to nicotine impairs antigen receptor-mediated signal transduction in lymphocytes. Toxicol App Pharmacol
4. Law R, Hackshaw K. Environmental tobacco smoke. Br Med Bull
5. Witschi H, Joad JP, Pinkerton KE. The toxicology of environmental tobacco smoke. Annu Rev Pharmacol Toxicol
6. Brondum J, Shu O, Steinbuch M, et al. Parental cigarette smoking and the risk of acute leukemia in children. Cancer
7. Sorahan T, Lancashire J, Hulten A, et al. Childhood cancer and parental use of tobacco: death from 1953 to 1955. Br J Cancer
8. Magnani C, Pastore G, Luzzatto L, et al. Parental occupation and other environmental risk factors in the etiology of leukemia and non-Hodgkin’s lymphoma in childhood: a case-control study [Abstract]. Tumori
9. Ji T, Shu O, Linet MS, et al. Parental cigarette smoking and the risk of childhood cancer among offspring of non smoking women. J Natl Cancer Inst
10. Weaver W, McAnally B, Bultery M, et al. Maternal smoking during pregnancy and childhood leukemia: finding of cancer and birth registry data sets. Am J Epidemiol
. 1997;145(11 suppl):49–56.
11. Shu O, Ross A, Pendergrass W, et al. Parental alcohol consumption, cigarette smoking and risk of infant leukemia: a children cancer group study. J Natl Cancer Inst
12. John M, Savitz A, Sandler DP. Prenatal exposure to parent’s smoking and childhood cancer. Am J Epidemiol
13. Johnson KC, Mao Y, Argo J, et al. The National Enhanced Cancer Surveillance System: a case-control approach to environment-related cancer surveillance in Canada. Environmetrics
14. Johnson KC, Pan S, Fry R, et al. Residential proximity to industrial plants and non-Hodgkin lymphoma. Epidemiology
15. Johnson KC, Hu J, Mao Y. Passive and active smoking and breast cancer risk in Canada, 1994–1997. Cancer Causes Control
16. Johnson KC, Hu J, Mao Y. Lifetime residential and workplace exposure to environmental tobacco smoke and lung cancer in never-smoking women, Canada 1994–97. Int J Cancer
17. Finkelstein MM. Leukemia after exposure to benzene: Temporal trends and implications for standards. Am J Ind Med
18. Darby SC, Nakashima E, Kato H. A parallel analysis of cancer mortality among atomic bomb survivors and patients with ankylosing spondylitis given X-ray therapy. J Natl Cancer Inst
19. Zeeb H, Blettner M. Adult leukemia: what is the rule of currently known risk factors? Radiat Environ Biophys
20. Greaves MF. Aetiology of acute leukemia. Lancet
21. Linet MS, Cartwright R. The leukemias. In: Schottenfeld D, Fraumeni JF, eds. Cancer Epidemiology and Prevention
. 2nd ed. New York: Oxford University Press; 1996:841–892.
22. Siemiatycki J, Nadon L, Lakhani R, et al. Exposure assessment. In: Siemiatycki J, ed. Risk Factors for Cancer in Workplace
. Boca Raton: CRC Press Inc.; 1991:45–115.
23. Breslow E, Day E. Statistical Methods in Cancer Research, Vol I: The Analysis of Case–Control Study
. Lyon, France: IARC Scientific Publication No. 32; 1980.
24. Curtin F, Morabia A, Bernstein M. Lifetime exposure to environmental tobacco smoke among urban women: differences by socio-economic class. Am J Epidemiol
25. Kreuzer M, Krauss M, Kreienbrock L, et al. Environmental tobacco smoke and lung cancer: a case-control study in Germany. Am J Epidemiol
26. Chelghoum Y, Danaila C, Belhabri A, et al. Influence of cigarette smoking on the presentation and course of acute myeloid leukemia. Ann Oncol
27. Miligi L, Seniori Costantini A, Crosignani P, et al. Occupational, environmental, and life-style factors associated with the risk of hematolymphopoietic malignancies in women. Am J Ind Med
28. Brown LM, Gibson R, Blair A, et al. Smoking and risk of leukemia. Am J Epidemiol
29. Tredaniel J, Boffetta P, Saracci R, Hirsch A. Exposure to environmental tobacco smoke and risk of lung cancer: the epidemiological evidence. Eur Respir
30. Henschen M, Frischer T, Pracht T. The internal dose of passive smoking at home depends on the size of the dwelling. Environ Res