Annals of Oncology Advance Access published online on January 22, 2008
Annals of Oncology, doi:10.1093/annonc/mdm578
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cigarette smoking and site-specific cancer mortality: testing uncertain associations using extended follow-up of the original Whitehall study
1 Medical Research Council Social and Public Health Sciences Unit, University of Glasgow, Glasgow
2 Department of Epidemiology and Public Health, University College London, London
3 Department of Social Medicine, University of Bristol, Bristol, UK
* Correspondence to: Dr G. D. Batty, Medical Research Council Social and Public Health Sciences Unit, University of Glasgow, 4 Lilybank Gardens, Glasgow G12 8RZ, UK. Tel: +44-141-357-7520; Fax: +44-141-337-2389; E-mail: david-b{at}msoc.mrc.gla.ac.uk
 |
Abstract |
|---|
 Top Abstract introductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
Background: The relation between cigarette smoking and several malignancies is still unclear. We examined the association of cigarette smoking with death attributed to 15 cancer sites, 7 of which are regarded as having an uncertain relation with tobacco.
Patients and methods: The original Whitehall study is a prospective cohort of 17 363 London-based male government employees (age 40–69 years) who were examined in the late 1960s and then followed up for a maximum of 38 years.
Results: Following adjustment for demographic characteristics, risk factors, and prevalent disease, established positive cigarette smoking—cancer gradients were confirmed for carcinoma of the lung, stomach, pancreas, bladder, upper aero-digestive (including oesophagus), and liver, and for myeloid leukaemia. Among the cancers of uncertain relation with smoking, mortality rates for malignancy of the colon, rectum and prostate and for lymphatic leukaemia were elevated in current and/or former smokers. There was essentially no apparent relation between smoking and mortality from carcinoma of the brain or from lymphoma.
Conclusion: In this study, cigarette smoking appears to be a risk factor for several malignancies of previously unclear association with tobacco use.
cancer, cigarette smoking, cohort study, Whitehall
|
introduction |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
The study of the association between tobacco smoking and cancer has a longer research tradition than is often recognised: case series accounts of a higher than expected occurrence of lip cancer among pipe smokers were made at least two centuries ago [1]. The first empirical investigation of risk appeared in the late 1930s when carcinoma of the lung was found to be more prevalent in men who smoked relative to their non-smoking counterparts [2]. Published later, Doll and Bradford Hill's work from the early 1950s is perhaps best known as identifying tobacco as a risk factor for lung cancer in case–control [3] and, subsequently, cohort studies [4].
More recently, cigarette smoking has also been shown to be related to an elevated risk of some less common but nonetheless important malignancies, including carcinoma of the stomach, pancreas, bladder, oesophagus, and kidney [5–8]. For several of these sites, the observational evidence, coupled with data from animal experimentation, is regarded as sufficiently secure for the effects of smoking to be described by some commentators and research agencies as causal, even for those tissues which do not fall into direct contact with tobacco smoke [5–8]. Despite apparent biological plausibility, uncertainty still exists as to the influence, if any, of smoking on other malignancies, including prostate, colon, rectum, brain, skin, and lymphoma. For many of these outcomes, this inconsistency of evidence may be at least partially ascribed to a paucity of sufficiently powered cohort studies.
In the Whitehall study, >19 000 male London-based government employees aged 40–69 years participated in a medical examination in the 1960s which included collection of a smoking history and a range of collateral data [9]. In extended (maximum 38 years) mortality surveillance of this cohort, there have been >3000 cancer deaths. These data afford us the rare opportunity of examining links between cigarette smoking and the afore-mentioned group of cancers for which a degree of uncertainty about the relation remains. Having extended mortality surveillance is of particular benefit when exploring the association between smoking and malignancies which are thought to have a particularly long induction period, such as cancer of the colon and rectum [10].
|
patients and methods |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
study population
In the Whitehall study, data were collected on 19 019 nonindustrial London-based male government employees aged between 40 and 69 years when examined between September 1967 and January 1970, representing a 74% response. This involved the completion of a study questionnaire and participation in a medical examination, both of which have been described in detail elsewhere [9]. In brief, the questionnaire included enquiries regarding civil service employment grade (an indicator of socioeconomic position [11]), intermittent claudication [12], angina, chronic bronchitis [13], marital status [14], physical activity [15], unexplained weight loss in the preceding year, and prescribed drug use. A series of questions pertaining to smoking were included in the questionnaire from which it was possible to determine whether or not the respondents had ever smoked, and, where applicable, the number of cigarettes smoked per day, and the duration of smoking (both for former and current smokers) [16]. Height, weight, forced vital capacity [17], ischaemia, fasting plasma cholesterol [18], postchallenge 2-h blood glucose [19], and blood pressure [20] were all determined using standard protocols which have been previously discussed. Diabetes was defined as blood glucose two hours after the glucose load of
11.1 mmol/l (200 mg/100 ml) and/or a positive response to the questionnaire enquiry "Are you, or have you been, diabetic?" Body mass index was computed using the usual formulae [weight (m)/height (kg2)] [21].
ascertainment of cancer mortality
Using the procedures of the National Health Service Central Registry, the records of 18 863 men (99.2% of cohort members) were traced and flagged until 30 September 2005. Among the 13 501 who died, 83.8% of death certificates were coded according to the eighth revision of the International Classification of Diseases (ICDs) [22], 6.2% according to the ninth [23] and 10.0% according to the tenth [24]. Codes for individual cancer sites in ascending ICD 8/9 code order were upper aero-digestive, including oesophagus (ICD-8/9: 140–150; ICD-10: C00–C15); stomach (ICD-8/9: 151; ICD-10: C16); colon (ICD-8/9: 153; ICD-10: C18); rectum (ICD-8/9: 154; ICD-10: C19); liver (ICD-8/9: 155–156; ICD-10: C22–C24); pancreas (ICD-8/9: 157; ICD-10: C25); trachea, bronchus, and lung (ICD-8/9: 162; ICD-10: C33–C34; referred to as ‘lung cancer’); melanoma (ICD-8/9: 172; ICD-10: C43); prostate (ICD-8/9: 185; ICD-10: C61); bladder (ICD-8/9: 188; ICD-10: C67); kidney (ICD-8/9: 189; ICD-10: C64–C66, C68); brain (ICD-8/9: 191; ICD-10: C71); lymphoma (ICD-8/9: 200–203; ICD-10: C81–C90); lymphatic leukaemia (ICD-8/9: 204; ICD-10: C91), and myeloid leukaemia (ICD-8: 205–206; ICD-9: 205–207; ICD-10: C91–C94).
Taking a similar approach to that used elsewhere [25], and on the basis of existing reviews [5–8], we divided cancer sites into a group which has been shown to have an established relation with cigarette smoking [eight malignancies: lung, stomach, pancreas, bladder, upper aero-digestive (including oesophagus), kidney, myeloid leukaemia, and liver] and a group for which this relationship may be regarded as uncertain (seven malignancies: colon, rectum, lymphoma, brain, melanoma, prostate, and lymphatic leukaemia), with an emphasis of our analytical efforts on the latter.
data analyses
A total of 17 985 men with mortality follow-up had data for smoking habit and all potential covariates. To simplify data interpretation, we dropped 622 men who reported smoking pipes or cigars. In addition, the cause of death for 41 men was unknown and they were also excluded, resulting in an analytical sample of 17 322 (91.3 % of those surveyed). Forty-nine former smokers and 30 current smokers with missing information on duration of tobacco consumption were also dropped but from specific analyses only.
We examined the relation of cancer risk with three smoking exposures: cigarette smoking status (never, former, and current); intensity (maximum number of cigarettes smoked per day for at least a year in former smokers and usual number of cigarettes smoked per day in current smokers); and duration (years of smoking in former and current smokers). The prevalence of baseline characteristics according to cigarette smoking status was adjusted for age (5-year age groups) using direct standardisation with the total study population as the standard. For continuous variables, least-squares means were used to present the age-adjusted means. Differences in the proportions or means between the never smokers and the former or current smokers were tested for significance by using a logistic regression (or least-squares regression for the continuous variables) adjusted for age and assessing the significance of the former and current smoking terms. Hazard ratios (HRs) and accompanying 95% confidence intervals (CIs) were computed for the relation of each of the smoking exposures with each cancer outcome using Cox proportional hazards regression model [26] with follow-up period as the time scale. In addition to age, adjustments were made for employment grade, physical activity, body mass index, marital status, systolic and diastolic blood pressure, cholesterol, forced expiratory volume, impaired glucose tolerance, diabetes, and disease at study entry. All statistical analyses were conducted using SAS computer software [27].
|
results |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
In Table 1, we show the relations between cigarette smoking status and study characteristics. In comparison to never smokers, currently smoking men were older, leaner, shorter, less active, of lower socioeconomic position, more likely to have a co-morbidity, and have lower lung function; conversely, current smokers had lower blood pressure levels. Former smokers generally had less favourable characteristics than never smokers.
|
Table 2 depicts relations between cigarette smoking and cancer sites in the Whitehall study which are widely regarded as established. In all analyses of the association between smoking and site-specific cancer mortality, differences in HRs between age-adjusted and multiply-adjusted models were small. Positive associations were apparent for cigarette smoking and each cancer mortality end point in Table 2, although, for some, statistical significance at conventional levels was not seen. Thus, current cigarette smoking was related to an elevated risk of carcinoma of the lung, stomach, bladder, myeloid leukaemia, and upper aero-digestive region (including oesophagus). Rates of death from pancreatic and liver cancer were also elevated in current smoking but CIs included unity. Incremental effects across the three smoking status groups (never, former, and current) were apparent for lung cancer and myeloid leukaemia. Carcinoma of the lung and upper aero-digestive region were positively related to number of cigarettes and duration of smoking (lung cancer only) in current smokers. Correlating cigarette smoking with all individual sites reported in Table 2 as a collective outcome resulted in a dose–response region effect across the smoking categories and a positive relation with number of cigarettes smoked per day and duration of smoking; a gradient that is likely to be generated by the very strong, and expected, smoking–lung cancer relationship.
|
When we examined the relation between smoking and noncancer outcomes (results not shown in detail, but available on request), as expected, cigarette smoking was associated with elevated rates of mortality from all causes (HRcurrent smoker versus never smoker; 95% CI 1.79; 1.70, 1.88), coronary heart disease (1.58; 1.44, 1.73), and stroke (1.28; 1.08, 1.51).
In Table 3, the relations between cigarette smoking and cancer sites in the Whitehall study which are generally regarded as uncertain are shown. In comparison to never smokers, being a former or current cigarette smoker at study induction was associated with an increased risk of death from carcinoma of the colon, rectum, and prostate. There was also some modest evidence that smoking was related to increased rates of mortality from lymphatic leukaemia, although not at conventional levels of statistical significance. Similarly, the suggestion that smoking conferred protection against melanoma—a stepwise gradient was seen—was, again, statistically nonsignificant. Once more, in analyses of the association between smoking and site-specific cancer mortality, differences in HRs between age-adjusted and multiply-adjusted models were modest.
|
In Table 4, the relation of cigarettes smoked and smoking duration in former and current smokers with malignancies with an uncertain association with this behaviour are shown. In both former and current smokers, number of cigarettes per day were positively related to rectal cancer. Rates of death from lymphatic leukaemia were also directly related to number of cigarettes smoked in current smokers, while other tobacco—cancer associations did not attain statistical significance.
|
|
discussion |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
The main aim of the present analyses was to examine the relation of cigarette smoking with death from cancers which have previously revealed uncertain relations with this behaviour. We also confirmed some previously established tobacco–malignancy associations. Of the uncertain smoking–cancer relations, there was evidence that rates of mortality from carcinoma of the colon, rectum, and prostate were raised in the current and/or former smokers. There was, however, essentially no apparent relation between smoking and brain cancer nor lymphoma.
The latest position statements and reviews [5–8] were unable to rule out bias and confounding as likely explanation for the apparent association of cigarette smoking with cancer of the colon and rectum. In the present analyses, which is based on an almost complete follow-up of men with a reasonably wide range of social, physiological, and behavioural covariates, we found elevated rates of colon cancer (in current smokers) and rectal cancer (in former smokers); however, neither were significantly related to number of cigarettes smoked. Temporal trends recently described by Giovannucci [10] are consistent with an induction period of three to four decades between genotoxic exposure to tobacco smoke and the diagnosis of colorectal cancer. Furthermore, cohort studies with extended follow-up which runs to 30+ years tend to reveal positive smoking—colorectal cancer associations which are generally not seen in cohorts with shorter surveillance periods [10].
plausible mechanisms
The mechanisms underlying the positive relations between smoking and mortality from some of the tumour sites featured in this report are likely to vary: cancer is not a single disease entity, as such there is unlikely to be a single unifying mechanism of tobacco-related carcinogenesis. The differential smoking—cancer relations by site may be at least partially ascribed to a differential duration of exposure to tobacco constituents and metabolites. For instance, the generally stronger relation of cigarette smoking with bladder cancer in comparison to cancer of the kidneys [5], which was apparent in the present study, may be due to the substantially longer storage time of urine in the bladder. A similar argument, however, did not find support in analyses of smoking in relation to colon and rectal cancer. Here, one would speculate that tobacco metabolite-carrying faeces that are in longer contact with the mucosa of the colon than the rectum—where they are only briefly held before evacuation—would lead to a stronger relation with carcinoma of the colon but this was not the case.
study limitations
While the Whitehall study has a number of strengths—not least its prospective design, extended follow-up, and large size which offers higher statistical power than many other studies in the field—it is not without its shortcomings. Firstly, while it is reasonably well characterised, the Whitehall study lacks data on potentially important covariates relevant to the aetiology of some tumour types. For instance, well-documented risk factors for stomach cancer, which also correlate with smoking habit, include Helicobacter pylori infection, a higher consumption of alcohol and salt-rich food, and, possibly, a lower intake of fruit and vegetables [28]. While we had data on dietary characteristics, including alcohol consumption, these were only available in a subgroup of study participants (N 
1500), which was too small to facilitate analyses with the cancer types featured in the current analyses. Secondly, smoking may have a differential impact on some cancers when division is made by anatomical or histological site; however, our end point data were taken from death certificates which, in the UK, do not offer this level of resolution. Thirdly, because the cancer outcomes herein were based on mortality, they therefore reflect a combination of incidence and survival rather than purely disease aetiology. Finally, because data on smoking habit were captured during a single baseline assessment rather than repeat measurement, there is likely to have been some exposure misclassification. Provided this misclassification is nondifferential with respect to the cancer of interest—which we believe it to be—it will have resulted in an underestimation of smoking—disease risk. Although the original Whitehall study participants were re-surveyed in older age [29], there are too few cancer deaths in men who provided information on smoking at both time points to facilitate analyses of these habits with subsequent tumour mortality rates.
In conclusion, in addition to established relations, in the present cohort, cigarette smoking was associated with elevated rates of cancer of the colon, rectum, and prostate. Despite a plethora of studies, some uncertainty remains over the relation of tobacco with selected cancer sites. Further examination of these links is therefore warranted.
|
funding |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
Department of Health and Social Security and the Tobacco Research Council; British Heart Foundation to MJS; Medical Research Council to MGM; Wellcome Trust Fellow to GDB (081021/Z/06/Z).
|
Acknowledgements |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
GDB generated the idea for the present manuscript and wrote the first draft around analyses conducted by MJS. The remaining authors commented on drafts of this manuscript including suggestions for further analyses. We thank Paul Brennan for his helpful comments. We also thank the civil servants who gave of their time to participate in the baseline survey in the 1960s.
Received for publication November 14, 2007. Revision received November 27, 2007. Accepted for publication November 29, 2007.
|
References |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionfundingAcknowledgementsReferences |
|---|
1. Redmond DE Jr. Tobacco and cancer: the first clinical report, 1761. N Engl J Med (1970) 282:18–23.[Web of Science][Medline]
2. Müller F. Tabakmissbrauch und lungencarcinoma. Z Krebsforsch (1939) 49:57–85.
3. Doll R, Bradford Hill A. Smoking and carcinoma of the lung; preliminary report. Br Med J (1950) 2:739–748.
4. Doll R, Bradford Hill A. A study of the aetiology of carcinoma of the lung. Br Med J (1952) 2:1271–1286.
5. Kuper H, Boffetta P, Adami HO. Tobacco use and cancer causation: association by tumour type. J Intern Med (2002) 252:206–224.[CrossRef][Web of Science][Medline]
6. Sasco AJ, Secretan MB, Straif K. Tobacco smoking and cancer: a brief review of recent epidemiological evidence. Lung Cancer (2004) 45(Suppl 2):S3–S9.
7. Vineis P, Alavanja M, Buffler P, et al. Tobacco and cancer: recent epidemiological evidence. J Natl Cancer Inst (2004) 96:99–106.
8. International Agency for Research on Cancer. Tobacco Smoke and Involuntary Smoking (IARC Monographs on the Evaluation of Carcinogenic Risk to Humans). IARC Monograph 83 (2004) Lyon: International Agency for Research on Cancer.
9. Reid DD, Hamilton PJS, McCartney P, et al. Cardiorespiratory disease and diabetes among middle-aged male civil servants. Lancet (1974) i:469–473.
10. Giovannucci E. An updated review of the epidemiological evidence that cigarette smoking increases risk of colorectal cancer. Cancer Epidemiol Biomarkers Prev (2001) 10:725–731.
11. Marmot MG, Shipley MJ, Rose G. Inequalities in death–specific explanations of a general pattern? Lancet (1984) 1:1003–1006.[CrossRef][Web of Science][Medline]
12. Davey Smith G, Shipley MJ, Rose G. Intermittent claudication, heart disease risk factors, and mortality. The Whitehall Study. Circulation (1990) 82:1925–1931.
13. Batty GD, Shipley MJ, Marmot MG, Davey Smith G. Leisure time physical activity and disease-specific mortality among men with chronic bronchitis: evidence from the Whitehall study. Am J Public Health (2003) 93:817–821.
14. Ben Shlomo Y, Davey Smith G, Shipley M, Marmot MG. Magnitude and causes of mortality differences between married and unmarried men. J Epidemiol Community Health (1993) 47:200–205.
15. Davey Smith G, Shipley MJ, Batty GD, et al. Physical activity and cause-specific mortality in the Whitehall study. Public Health (2000) 114:308–315.[Web of Science][Medline]
16. Ben Shlomo Y, Smith GD, Shipley MJ, Marmot MG. What determines mortality risk in male former cigarette smokers? Am J Public Health (1994) 84:1235–1242.
17. Batty GD, Gunnell D, Langenberg C, et al. Adult height and lung function as markers of life course exposures: associations with risk factors and cause-specific mortality. Eur J Epidemiol (2006) 21:795–801.[CrossRef][Web of Science][Medline]
18. Davey Smith G, Shipley MJ, Marmot MG, Rose G. Plasma cholesterol concentration and mortality. The Whitehall Study. JAMA (1992) 267:70–76.
19. Fuller JH, Shipley MJ, Rose G, et al. Coronary-heart-disease risk and impaired glucose tolerance. The Whitehall study. Lancet (1980) 1:1373–1376.[Web of Science][Medline]
20. Batty GD, Shipley MJ, Marmot MG, Davey Smith G. Blood pressure and site-specific cancer mortality: evidence from the original Whitehall study. Br J Cancer (2003) 89:1243–1247.[CrossRef][Web of Science][Medline]
21. Batty GD, Kivimaki M, Smith GD, et al. Obesity and overweight in relation to mortality in men with and without type 2 diabetes/impaired glucose tolerance: the original Whitehall study. Diabetes Care (2007).
22. World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death (Eighth Revision) (1967) Geneva: World Health Organisation.
23. World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death (Ninth Revision) (1977) Geneva: World Health Organization.
24. World Health Organization. International Statistical Classification of Diseases and Related Health Problems (10th revision) (1992) Geneva: World Health Organization.
25. Doll R, Peto R, Boreham J, Sutherland I. Mortality from cancer in relation to smoking: 50 years observations on British doctors. Br .J Cancer (2005) 92:426–429.[Web of Science][Medline]
26. Cox DR. Regression models and life-tables. J R Stat Soc Ser B (1972) 34:187–220.
27. SAS Institute Inc. SAS/STAT(R) User's Guide, Version 6, Volumes 1 & 2 (1989) Cary, NC: SAS Institute Inc.
28. Kelley JR, Duggan JM. Gastric cancer epidemiology and risk factors. J Clin Epidemiol (2003) 56:1–9.[CrossRef][Web of Science][Medline]
29. Breeze E, Fletcher AE, Leon DA, et al. Do socioeconomic disadvantages persist into old age? Self-reported morbidity in a 29-year follow-up of the Whitehall Study. Am J Public Health (2001) 91:277–283.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. L. Watters, Y. Park, A. Hollenbeck, A. Schatzkin, and D. Albanes Cigarette Smoking and Prostate Cancer in a Prospective US Cohort Study Cancer Epidemiol. Biomarkers Prev., September 1, 2009; 18(9): 2427 - 2435. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Botteri, S. Iodice, V. Bagnardi, S. Raimondi, A. B. Lowenfels, and P. Maisonneuve Smoking and Colorectal Cancer: A Meta-analysis JAMA, December 17, 2008; 300(23): 2765 - 2778. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||