Cancer Causes and Control 12: 551±556, 2001. Ó 2001 Kluwer Academic Publishers. Printed in the Netherlands.
Cigar, pipe, and cigarette smoking and bladder cancer risk in European men A. Pitard1, P. Brennan1, J. Clavel2, E. Greiser3, G. Lopez-Abente4, J. Chang-Claude5, J. Wahrendorf5, C. Serra6, M. Kogevinas7 & P. Boetta1,* 1 Unit of Environmental Cancer Epidemiology, International Agency for Research on Cancer, 150 cours AlbertThomas, 69008 Lyon, France; Ph.: +33-4-72738441; Fax: +33-4-72738342; E-mail: bo[email protected]
; 2Unit 170, National Institute of Health and Medical Research (INSERM), Villejuif, France; 3Bremen Institute for Prevention Research and Social Medicine, Bremen, Germany; 4Unit of Cancer Epidemiology, National Centre for Epidemiology, Madrid, Spain; 5German Cancer Research Centre, Heidelberg, Germany; 6Center for Studies, Medical Programs and Services, Parc Tauli Foundation, Sadabell, Spain; 7Municipal Institute of Medical Investigation, Barcelona, Spain (*Author for correspondence) Received 7 August 2000; accepted in revised form 14 February 2001
Key words: bladder neoplasms, cigar, cigarettes, epidemiology, pipe.
Abstract Objective: Estimating the risk of bladder cancer from cigar and pipe smoking is complicated by a small number of non-cigarette smokers included in most relevant studies. Methods: We undertook a pooled analysis of the data on men from six published case±control studies from Denmark, France, Germany, and Spain, to assess the association between pipe and cigar smoking and bladder cancer, and to compare it with the risk from cigarette smoking. Complete history of tobacco smoking was ascertained separately for cigarettes, cigars, and pipe. Odds ratios (ORs) were estimated after adjusting for age, study, and employment in high-risk occupations. Results: The pooled data set comprised 2279 cases and 5268 controls, of whom 88 cases and 253 controls smoked only cigars or pipe. The OR for pure cigarette smoking was 3.5 (95% con®dence interval [CI] 2.9±4.2), that for pure pipe smoking was 1.9 (95% CI 1.2±3.1) and that for pure cigar smoking was 2.3 (95% CI 1.6±3.5). The increase in the OR of bladder cancer that was observed with duration of smoking was non-signi®cantly lower for cigars than for cigarettes. Conclusion: Our results suggest that smoking of cigars and pipe is carcinogenic to the urinary bladder, although the potency might be lower than for cigarettes.
Introduction The causal association between cigarette smoking and bladder cancer risk is well established [1, 2]. However, information on a carcinogenic eect of other tobacco products on the bladder is limited and con¯icting. Indeed, most relevant epidemiological studies include relatively small numbers of persons who are exclusively pipe or cigar smokers, and only a few studies include a suciently large number of pure cigar and pipe smokers to allow for quantitative analysis [3±9]. In a large study from the United States, pure pipe smokers were estimated to have an odds ratio (OR) of 1.2 (95% con®dence interval [CI] 0.75±2.0) and
pure cigar smokers were estimated to have an OR of 1.3 (95% CI 0.92±1.9) . A higher relative risk was observed for those who inhaled pipe smoke deeply (OR 3.1, 95% CI 1.3±7.5). In a further analysis of a subset of data from the same study, an increased risk of bladder cancer (OR 2.5, CI 1.0±6.0) was seen for pure cigar smokers, but no elevated risk was observed for pipe smokers . Morrison et al.  conducted a large international case±control study in Boston (USA), Manchester (UK), and Nagoya (Japan). They found that, among men from Manchester who had never smoked cigarettes, the relative risk of bladder cancer associated with ever smoking a pipe was 3.9 (95% CI 1.3±11.8). In the remaining studies [3, 7±9], cigar and
A. Pitard et al.
pipe smoking was not associated with risk of bladder cancer. We have recently reported that cigar and pipe smoking among European men might entail a risk of lung cancer that is higher than comparable estimates obtained in American populations . We have therefore undertaken a similar study of bladder cancer risk from cigar and pipe smoking in Europe, a region for which limited information is available [3, 4].
assessed by ascertaining if subjects were ever employed in any of the high-risk occupations, as described by 't Mannetje et al. . OR values and 95% CIs were estimated by unconditional logistic regression after adjusting for age (5 years), study, and employment in high-risk occupations (yes or no). All statistical analyses were performed using the Stata software .
Materials and methods
A total of 154 cases and 1109 controls were classi®ed as non-smokers. Pure cigarette, pure cigar, pure pipe, and mixed smokers represent respectively 57.2%, 2.3%, 1.5%, and 22.3% of all smokers (Table 2). Most mixed smokers smoked cigarettes in combination with other products. Transitional cell carcinoma was the most common histological type, representing over 92% of con®rmed histologies. A total of 88 cases and 253 controls smoked only cigars or pipe. The relative risk of bladder cancer for smoking of any tobacco product versus non-smokers was 3.5 (95% CI 2.9±4.2). The study-speci®c ORs varied from 2.6 to 4.2, with the exception of one study in which a higher OR was reported (Table 1). The results for smokers of dierent tobacco products are shown in Table 2. The risk was higher for pure cigarette smokers than for pure pipe, pure cigar, and mixed smokers. An increase in risk of bladder cancer was observed with increasing duration of smoking of any tobacco product (Table 3). However, the slope of the dose±response relation was stronger for cigarette smoking than for cigar and pipe smoking. A protective eect of time since quitting was also present for all types of smoking: OR was compared to current smokers, relative risk was 0.3 (95% CI 0.2±0.8) among former cigar smokers (10 or more years since quitting), 0.4 (95% CI 0.2±1.0) among former pipe smokers, and 0.3 (95% CI 0.5±0.8) among former cigarette smokers. The analysis of daily tobacco consumption was restricted to two studies [14, 15] (Table 4). A dose± response relation was observed between risk of bladder cancer and number of cigarettes smoked per day, which was less steep after adjustment for duration of smoking. For cigars, the increase in risk was imprecise but comparable to that found for cigarettes. There were too few pure pipe smokers with information on their amount of smoking to allow for a meaningful analysis.
We undertook a pooled analysis of the data from six published case±control studies to assess the association between the use of pipes and cigars and the risk of bladder cancer, and compared it with the risk from cigarette smoking. Studies were from Denmark, France, plus two each from Germany and Spain [11±16]. We excluded: (a) women, because of the small number of cigar or pipe smokers among them; (b) cases without histologically con®rmed malignant tumor of the bladder; (c) prevalent cases (de®ned as those with more than 2 years time between diagnosis and interview); and (d) subjects aged 80 or older. The pooled data set comprised a total of 2279 cases and 5268 controls. Selected characteristics of the design of the participating studies are presented in Table 1. The German studies contributed 48% of subjects, the Spanish studies 24%, the French study 16%, and the Danish study 12%. All cases were recruited in urological hospital departments. Controls were hospital patients in two studies (mainly from urological departments), a sample of the underlying population in three studies, and mixed hospital and population controls in one study. Patients hospitalized for tobacco-related conditions (e.g. cardiovascular and respiratory diseases) were excluded from the pool of eligible hospital controls (Table 1). The complete history of tobacco smoking was ascertained through intervieweradministered questionnaires separately for cigarettes, cigars or cigarillos (the latter were not distinguished in the questionnaire, and were classi®ed thereafter as cigars), and for pipe. All subjects were assigned to one of the following mutually exclusive smoking categories: non-smokers, pure cigarette smokers, pure cigar smokers, pure pipe smokers, cigarette smokers who also smoked cigars or pipe, and mixed smokers of cigars and pipe (but not cigarettes). Duration of smoking and time since quitting (for ex-smokers) were ascertained in all studies. Average daily consumption of cigarettes, cigars, and pipe was ascertained in two studies [14, 15]. We did not attempt to estimate cumulative exposure to tobacco. Occupational exposure to bladder carcinogens was
Discussion We have found a carcinogenic eect of cigar or pipe smoking on the urinary bladder, which is non-signi®cantly
Barcelona, Madrid, Cadiz, Quipuzcoa, Vizcaya
Berlin West, Leverkusen, Bremen, Halle, Merseburg, Jena
Paris, Lille, Strasbourg
Source of cases
Response rate (%) cases/controls
(1) Patients admitted in the same hospital for causes other than respiratory, circulatory and urinary diseases. (2) Census of municipal register of the case Residents of municipalities of the cases
Urology wards and homes for elderly in the same town Residents of the municipalities of cases
Patients admitted in the same hospital for causes other than cancer
Residents of the municipalities of cases
Sources of controls
a Odds ratio of bladder cancer and 95% con®dence interval for ever smoking of any tobacco product, adjusted for age and gender. NA; not available.
Mean year of enrolment
Table 1. Studies included in the pooled analysis
Ever smoking: (95% CI)a
Smoking and bladder cancer risk 553
A. Pitard et al.
Table 2. Odds ratio of bladder cancer for smoking of dierent tobacco products Type of smokers
No. of cases
No. of controls
95% con®dence interval
Never smokers Pure cigarette smokers Cigarette (and cigar or pipe) smokers Pure pipe smokers Pure cigar smokers Cigar and pipe smokers
154 1420 617 28 50 10
1109 2895 1011 85 122 46
1 3.5 3.7 1.9 2.3 1.3
Referent category 2.9±4.2 3.1±4.6 1.2±3.1 1.6±3.5 0.6±2.6
Adjusted by age, center, and occupational exposures.
Table 3. Risk of bladder cancer and duration of cigar, pipe, and cigarette smokinga Odds ratiob
Years of smoking
No. of cases
No. of controls
Non-smokers Pure cigar smokers 1±29 30±39 ³40 Test for linear trend, p-value
15 12 22
62 28 32
1.4 2.7 3.8