Tobacco smoke and bladder cancer—in the European prospective investigation into cancer and nutrition
The purpose of the present study was to investigate the association between smoking and the development of bladder cancer. The study population consisted of 429,906 persons participating in the European Prospective Investigation into Cancer and Nutrition (EPIC), 633 of whom developed bladder cancer during the follow-up period. An increased risk of bladder cancer was found for both current- (incidence rate ratio 3.96, 95% confidence interval: 3.07–5.09) and ex- (2.25, 1.74–2.91) smokers, compared to never-smokers. A positive association with intensity (per 5 cigarettes) was found among current-smokers (1.18, 1.09–1.28). Associations (per 5 years) were observed for duration (1.14, 1.08–1.21), later age at start (0.75, 0.66–0.85) and longer time since quitting (0.92, 0.86–0.98). Exposure to environmental tobacco smoke (ETS) during childhood increased the risk of bladder cancer (1.38, 1.00–1.90), whereas for ETS exposure as adult no effect was detected. The present study confirms the strong association between smoking and bladder cancer. The indication of a higher risk of bladder cancer for those who start smoking at a young age and for those exposed to ETS during childhood adds to the body of evidence suggesting that children are more sensitive to carcinogens than adults. © 2006 Wiley-Liss, Inc.
Bladder cancer is the fourth most common malignancy among Western men.1 It is ∼3.5 times more frequent among men than among women.2 About 71% of all bladder cancer patients are older than 65 years.2
Several studies have shown an association between cigarette smoking and bladder cancer.3, 4, 5, 6, 7, 8 Overall, ever-smokers appear to have a two to three times higher risk of bladder cancer than never-smokers. All time-related smoking variables–duration, time since quitting and age at start of smoking—have been shown to affect the development of bladder cancer when studied individually,3, 4, 5, 6, 7, 8 but only few studies have investigated independent, mutually adjusted effects. The apparent higher risk associated with an early start has been ascribed to a longer duration for those who start at a young age as opposed to an independent effect of exposure early in life,4 but there is growing evidence that children and adolescents may be more sensitive to carcinogens than adults.9, 10 Studies on smoking intensity have shown a dose-response for current smokers and for ex- and current smokers combined in relation to bladder cancer,3, 4, 5, 6, 7, 8 but the effect of smoking intensity for ex-smokers alone has not been investigated. In several studies a levelling-off of the effect of intensity and duration has been found.11
It has also been shown that intake of fruit and vegetables may have a protective effect on bladder cancer.12, 13
Both exposure to environmental tobacco smoke (ETS) as adult and during childhood have been shown to increase the risk of lung cancer.5, 14 For bladder cancer information is sparse: one prospective cohort study reported a nonsignificant positive association between bladder cancer, and respectively, ETS exposure in childhood and ETS exposure at work,3 and two studies reported no effect of ETS exposure in adults,15, 16 but the possible association between ETS exposure and bladder cancer needs further investigation.
We used the large European Prospective Investigation into Cancer and Nutrition (EPIC) cohort to estimate mutually adjusted effects on bladder cancer of average lifetime intensity, duration, time since quitting and age at start of cigarette smoking and of exposure to ETS.
Participants and methods
EPIC is a multicentre prospective cohort study consisting of 23 centres from Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden and the United Kingdom. The methods have been reported in full by Riboli et al.17 In brief, the study populations were aged 25 to 70 years at the time of recruitment and were mostly recruited from the general population residing in a specific geographical region. Eligible subjects were invited to participate in the study, and those who accepted gave informed consent and completed lifestyle questionnaires. The lifestyle questionnaires included questions on education and socioeconomic status, occupation, previous illness, alcohol and tobacco consumption and physical activity. In addition, in many centres, questions of exposure to ETS were included. In most centres, diet was measured by country-specific, self-administered questionnaires, though some used interviewers.
Follow-up was based on population cancer registries in seven of the participating countries: Denmark, Italy, the Netherlands, Norway, Spain, Sweden and the United Kingdom. In France, Germany and Greece a combination of methods were used including health insurance records, cancer and pathology registries, and active follow-up through participants and their next-of-kin. Mortality data were also obtained from either the cancer registry or mortality registries at the regional or national level. For the current analysis participants were followed from study entry (1991–2000) until a primary bladder cancer diagnosis (ICD-10 code C67), death, emigration or end of the follow-up period (2000–2004). All urothelial neoplasms were included whether they were registered as pTa, pTis or pT1+. Only patients with urothelial cell carcinomas (=transitional cell carcinomas) were included in the analyses. Twelve patients with a squamous cell carcinoma, adenocarcinoma or another rare morphology were excluded.
Most centres collected very detailed lifetime history of tobacco consumption: smoking status at inclusion (current, ex, never), type of tobacco (cigarettes, cigars, pipe), amount of tobacco smoked, duration of smoking at inclusion and age at start, and quitting of smoking. Average daily cigarette smoking intensity was calculated as a lifetime mean value from self-reported information on cigarette smoking intensities at ages 20–29, 30–39, 40–49 and 50–59. The questionnaires used for data collection were not very detailed concerning ETS exposure and were not identical in the included countries. We, therefore, used broad ETS exposure categories knowing that minor differences existed within these categories: “Exposed as adult” were participants stating at the time of recruitment to be exposed to ETS at home and/or at work, and “exposed in childhood” were participants stating that their parents smoked or that they were otherwise exposed to ETS during childhood.
In the present study, the centres in Umeå (24,811 participants) and Naples (5,058 participants) were excluded because of missing information on both average lifetime smoking intensity and ETS exposure. Furthermore, participants with missing information on diet (7,042) or smoking characteristics (26,737) were excluded. The analyses of cigarette smoking were based on data from 337,934 cohort members from the 19 centres with information on average lifetime intensity of cigarette smoking. The analyses of ETS exposure were based on information from 126,908 participants from the four centres with information on ETS exposure both in childhood and as adult.
The analyses were based on Cox proportional hazard models, in which age was defined as the follow-up variable. Through this, an optimal adjustment for the possible confounding effect of age is obtained. Time under study (follow-up) was included as a time-dependent variable and was modelled as a broken line with breaks at 1 year, 3 years, and 5 years after entry into the cohort study.18 We also analysed the smoking variables for possible effect modification caused by different follow up times, in a model where participants were categorized according to follow-up time, but did not find any.
All models were stratified by gender and country and unless otherwise stated, models were further adjusted for baseline values of fruit (linear) and vegetable (linear) consumption. Two-sided 95% confidence intervals (CI) for the incidence rate ratio (IRR) were calculated based on Wald's test of the Cox regression parameter, that is, on the log rate ratio scale.
The effects of intensity, duration, age at start and time since quitting of smoking, were calculated both as risk estimates for categories of the variables and as linear risk estimates. The hypothesis of a linear association was evaluated using a linear spline with three boundaries, placed at the quartiles among cases (for maximum power), as covariates in the Cox model.18 The linearity was evaluated graphically as well as by a numerical test using the likelihood ratio test statistic to compare the model assuming linearity with the linear spline model. We found no statistically significant deviation from linearity (all p > 0.06). Since previous studies have shown a nonlinear dose-response pattern of intensity and duration, analyses were performed both assuming linearity of duration and intensity and with duration and intensity modelled as linear splines with 1 break and with 3 breaks, wherever adjusting for these variables.
Possible effect modification of smoking intensity by smoking duration, age at start and time since quitting was analysed by comparing the linear effect of intensity for the four quartiles of each time related variable.
The average follow up time was 6.3 years. Overall, 72% of the included participants were female, contributing 30% of the bladder cancer cases. Relatively more cases than cohort members were exposed to ETS during childhood, both among never-smokers and ever-smokers (Table I). On average, cases consumed slightly less fruit and vegetables than cohort members (Table I). At baseline, cases had on average smoked more than ten years longer than cohort members, they had a higher smoking intensity and they had started smoking earlier, whereas the number of years since quitting for ex-smokers was similar among cases and cohort members (Table I).
Table I. Baseline Characteristics (At Time of Recruitment)
| Male (%)||40||18||77||39|
| Female (%)||60||82||23||61|
| Exposed as child (%)||83||68||85||80|
| Exposed as adult (%)||62||63||88||81|
|Highest school level|
| None (%)||4||6||3||3|
| Primary (%)||28||22||40||25|
| Technical/professional (%)||26||19||26||27|
| Secondary school (%)||14||24||13||21|
| University degree (%)||25||25||17||22|
|Length of follow-up (years)||4 (0–8)2||7 (3–9)2||4 (0–9)2||6 (3–9)2|
|Age (years)||59 (45–73)||52 (33–67)||60 (49–71)||51 (36–66)|
| Fruit intake||224 (55–591)||232 (47–648)||140 (19–506)||180 (26–595)|
| Vegetable intake||170 (49–435)||196 (64–508)||131 (39–375)||162 (53–484)|
| Duration of smoking (years)||–||–||37 (10–50)||24 (4–44)|
| Time since quitting (years)3||–||–||14 (2–37)||14 (2–34)|
| Age at start (years)||–||–||17 (13–27)||18 (14–30)|
| Average intensity of smoking (cig/day)4||–||–||15 (3–30)||12 (2–28)|
| Current intensity of smoking (cig/day)5||–||–||17 (3–31)||15 (2–30)|
Adjustment for highest obtained school level did not change results (data not shown). Adjustment for having ever been employed in a high bladder cancer risk job, for the centres where information on occupation was available, did not change results (data not shown).
We found no statistically significant differences among countries (all p > 0.41) or genders (all p > 0.12) in the relation between bladder cancer and any of the smoking characteristics, though women had a 31% increased risk of bladder cancer pr 5 cigarettes/day (IRR: 1.31 95% CI: 1.11–1.55) and men only had a 13% increased risk pr 5 cigarettes/day (IRR: 1.13 95% CI: 1.03–1.24).
Compared to never-smokers current-smokers had a four times and ex-smokers a two times higher risk of developing bladder cancer (Table II). For current-smokers the results showed a significant increase in bladder cancer risk with increasing smoking intensity, whereas among ex-smokers the association was weaker (Table II). Among ex-smokers an increase in risk with increasing smoking intensity was only seen for those who had quit smoking 1–7 years earlier (data not shown). For those who quit more than 7 years earlier increasing intensity did not affect the risk level. No effect modifications of smoking intensity were detected by duration or age at start (data not shown). Further, Table II shows that longer duration of smoking increases and longer time since quitting smoking and later age at starting smoking decreases the risk of bladder cancer.
Table II. Effect of Different Smoking Variables, in Categories, on the Risk for Bladder Cancer1
|Average intensity (cig/day)|
| Current smokers|
| Ever smokers|
|Time since quit (years)|
|Age at start (years)|
Table III shows the linear effects of the smoking characteristics calculated per 5 years or 5 cigarettes/day. The associations of the characteristics related to timing of smoking were maintained when adjusted for intensity (model 2), but when mutually adjusted, all effect estimates were lowered (model 3) and only age at start remained statistically significant with a 19% reduction in risk for every 5 years later start. The effects were virtually unchanged in analyses with adjustment for intensity and duration modelled as nonlinear variables (data not shown).
Table III. Linear Effect, Per 5 Cigarettes/Day OR 5 Years1, of Smoking Variables on the Risk for Bladder Cancer with Varying Adjustments2
|Average intensity (per 5 cig/day)|
| Ex-smokers||1.06||0.99–1.15||0.11|| || || ||1.04||0.96–1.12||0.35|
| Current smokers||1.18||1.09–1.28||<0.0001|| || || ||1.16||1.07–1.26||0.0004|
|Duration (per 5 years)|
| Ever smokers||1.14||1.08–1.21||<0.0001||1.13||1.07–1.20||<0.0001||1.05||0.93–1.19||0.44|
|Time since quitting (per 5 years)|
| Ex smokers||0.92||0.86–0.98||0.01||0.92||0.86–0.99||0.02||0.96||0.84–1.10||0.53|
|Age at start (per 5 years)|
| Ever smokers||0.75||0.66–0.85||<0.0001||0.78||0.69–0.88||<0.0001||0.81||0.68–0.96||0.02|
Results in Table IV were based on a subsample of the total population with information on ETS exposure. Risk estimates for ex- (IRR: 2.69, 95% CI: 1.90–3.80) and current-smokers (IRR: 4.03, 95% CI: 2.87–5.65) in the subpopulation were similar to those observed for the total population. The results shown in Table IV indicate no association between exposure to ETS as adult and the development of bladder cancer. Exposure to ETS during childhood was associated with an almost 40% increased risk of bladder cancer compared to no such exposure. No statistically significant difference between estimates for never-, ex and current-smokers was detected (p = 0.38).
Table IV. The Risk for Bladder Cancer in Persons Exposed to Environmental Tobacco Smoke Compared to Persons not Exposed to Environmental Tobacco Smoke1
|In adulthood (Exposed as adult vs. not exposed as adult)|
|In childhood (Exposed as child vs. not exposed as child)|
The present study confirms the strong association between smoking and bladder cancer, and the benefit of quitting. The study indicates that children and adolescents may be more susceptible to tobacco carcinogens than adults, as evidenced by a higher risk for those who start smoking at a young age and for those exposed to ETS during childhood. For ex-smokers the past smoking intensity was not associated with the development of bladder cancer.
In addition to the large number of cases, one strength of the present study was the prospective design with low potential for recall or selection bias. The participants were asked about early smoking habits and ETS exposure in childhood, when included in the study at around 50–60 years of age, which might have caused misclassification in the exposure assessment.
The development of bladder cancer was associated with all investigated smoking characteristics: intensity, duration, time since quitting and age at start. Further, the results indicate the existence of a harmful effect of starting smoking at a young age, which is not associated with duration of smoking. However, duration, time since quitting and age at start of smoking are correlated; and the mutually adjusted results should be treated with caution.
Our study indicates that exposure to ETS might increase the risk of bladder cancer for those exposed in childhood, but not for those exposed as adults, which is concordant with the results of the three previous studies which have explored the association between bladder cancer and ETS exposure during childhood3 and as adult.3, 15, 16
A greater sensitivity of children than of adults to the bladder carcinogenic effect of ETS exposure is also in line with our finding indicating that starting to smoke at a young age increases the risk of bladder cancer. Furthermore, molecular and epidemiologic evidence indicate that children are more vulnerable to environmental carcinogens than adults.9, 10 The indication of a bladder carcinogenic effect of ETS exposure during childhood should be treated with caution due to the small numbers of cases and the result should be confirmed in future studies.
In our study, average lifetime smoking intensity had a strong positive association with bladder cancer risk among current smokers, but not among long-time ex-smokers, indicating that heavy smokers, in particular, will be able to reduce their risk of bladder cancer substantially if they quit smoking.
The indication in our study that early exposure to tobacco smoke might increase the risk of bladder cancer calls for further research and adds to the body of evidence suggesting that children are more sensitive to carcinogens than adults.
We thank all the participants in EPIC