Cigarette Smoking, Smoking Cessation, and Bladder Cancer Risk: A Pooled Analysis of 10 Cohort Studies in Japan

Background Although cigarette smoking is an established risk factor for bladder cancer, assessment of smoking impact on bladder cancer in Asian populations has been hindered by few cohort studies conducted in Asian populations. Therefore, we investigated the risk of bladder cancer associated with smoking status, cumulative smoking intensity, and smoking cessation in Japan. Methods We analyzed data for 157,295 men and 183,202 women in 10 population-based cohort studies in Japan. The risk associated with smoking behaviors was estimated using Cox regression models within each study, and pooled hazard ratios (HRs) and their 95% confidence intervals (CIs) for the incidence of bladder cancer were calculated. Results During 4,729,073 person-years of follow-up, 936 men and 325 women developed bladder cancer. In men, former smokers (HR 1.47; 95% CI, 1.18–1.82) and current smokers (HR 1.96; 95% CI, 1.62–2.38) had higher risk than never smokers. In women, current smokers had higher risk than never smokers (HR 2.35; 95% CI, 1.67–3.32). HRs in men linearly increased with increasing pack-years. Risk decreased with increasing years of smoking cessation in men, with a significant dose-response trend. Former smokers with a duration of more than 10 years after smoking cessation had no significantly increased risk compared with never smokers (HR 1.26; 95% CI, 0.97–1.63). Conclusion Data from a pooled analysis of 10 population-based cohort studies in Japan clearly show an association between cigarette smoking and bladder cancer risk. The risk of smokers may approximate that of never smokers following cessation for many years.


INTRODUCTION
Incidence rates of bladder cancer (BCa) differ among regions, and are the highest in Europe. 1 In Central and Eastern Asia, where incidence is relatively low, the highest incidence of BCa is found in Japanese men. 1 Despite a significant decrease in incidence from 2003 through 2015, the mortality of BCa in Japan has been stable. 2Improving mortality may require not only the development of effective treatments for advanced BCa but also stronger primary prevention to decrease BCa incidence; in particular, primary prevention requires accurate evaluation of risk factors for BCa among ethnic groups.
Cigarette smoking is an established modifiable risk factor for BCa, 3 with a population-attributable risk of approximately 50%. 4 Previous meta-analyses demonstrating an association of BCa risk with smoking were mainly derived from Western populations. 5,6hese previous meta-analyses indicated that the risk of BCa associated with smoking was lower in Asian populations than in either European or North American populations.][9][10] To accurately investigate this risk in Japan, we previously conducted a systematic review of three cohort and eight case-control studies in Japanese populations. 11Results showed a summary relative risk of 2.14 among ever smokers compared with never smokers, 11 which also suggests a smaller impact of smoking on BCa risk in Japanese than Western populations.
Nevertheless, because these studies categorized smoking differently and did not all provide detailed information on smoking habits, we were unable to examine the association between cumulative exposure to smoking intensity and BCa risk.In contrast, analyses of data collected by large-scale prospective studies can assess the impact of smoking using the same cut-points for cumulative smoking intensity and duration.In addition, few cohort studies in Asian populations have evaluated the impact of smoking cessation on BCa risk, although doing so could encourage policymakers and health care workers to re-emphasize the importance of counselling for patients to quit smoking.
Here, to investigate the association between BCa risk and smoking behaviors, such as smoking status, cumulative smoking intensity, and smoking cessation, we conducted a pooled analysis of 10 population-based cohort studies in Japan.

Study populations
Participant data were collected from 10 major populationbased cohort studies in Japan, namely the Japan Public Health Center-based Prospective Study (JPHC-I and -II), 12 the Japan Collaborative Cohort Study (JACC), 13 the Miyagi Cohort Study (MIYAGI-I), 14 the Three-Prefecture Cohort Study in Miyagi (MIYAGI-II), 15 the Three-Prefecture Cohort Study in Aichi (AICHI), 15 the Three-Prefecture Cohort Study in Osaka (Osaka), 15 the Takayama Study (TAKAYAMA), 16 the Ohsaki Cohort Study (OHSAKI) 17 and the Life-Span Study (LSS). 18hese cohort studies were selected in accordance with the inclusion criteria, as follows: (i) started in the mid-1980s to mid-1990s in Japan, (ii) included >30,000 participants, (iii) collected self-reported data on smoking habits, and (iv) confirmed incidence of all cancers during the follow-up period.We excluded participants who had a history of other primary cancer, had missing information on smoking, or were exposed to atomic bomb radiation of ≥100 mGy (for LSS).Selected characteristics of the participants in each study are shown in Table 1.Written informed consent was obtained from all participants in each study, and the study protocols were approved by the institutional review boards of respective study centers.

Assessment of smoking
Each of the studies collected information on smoking behavior using self-administered questionnaires at baseline.The information included smoking status, age of smoking initiation, number of cigarettes smoked per day, and age at quitting smoking.Smoking status was categorized into never, former, and current smoking.Cumulative smoking intensity was estimated in pack-years (PY), defined as the number of cigarettes smoked per day divided by 20 (a pack) multiplied by the number of years of smoking, and categorized into 0, ≤20, 20-40, or >40 PY.Years since smoking cessation were categorized into ≥10, 5-9, or 0-4 among former smokers.We restricted analyses assessing the effects of cumulative smoking intensity and smoking cessation to men because of the small number of experienced smokers among women.

Follow-up and study outcome
Participants were followed from the baseline survey until the last date of follow-up in each study.Study outcome was defined as the incidence of BCa newly diagnosed during the study period.New-onset BCa cases were identified through population-based cancer registries and/or active patient notification from major local hospitals.BCa was coded according to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) or the International Classification of Diseases and Health Related Problems, Tenth Revision (ICD-10).

Statistical analysis
Person-years of follow-up were counted from the date of enrollment in each study to the date of BCa diagnosis, migration from the study area, death from any cause or end of follow-up, whichever came first.Cox regression models were applied to calculate hazard ratios (HRs) and their 95% confidence intervals (CIs) for the incidence of BCa.The following two models were established: model 1 adjusted for age (continuous) and area (applicable to JPHC, JACC and LSS only), and model 2 excluded cases diagnosed within the first 3 years from enrollment to minimize reversal of cause and effect.We first calculated the study-specific HRs and then applied random-effects models to obtain pooled HR estimates.When any study identified no BCa cases for a category, we pooled HRs for the corresponding category by excluding data in that study.Dose-response relationships were assessed by incorporating PYs or years since smoking cessation into the models as continuous variables, providing HRs per increase in 1 PY or 1 year of smoking cessation (unit HRs).The JPHC study and one pooled study previously reported an association between BCa risk and smoking behaviors. 8,19Therefore, we updated the dataset to allow longer follow-up.Heterogeneity across studies for each category was assessed using Cochran's Q-statistic and I 2 -statistic.Statistical analyses were performed using STATA 15 (Stata Statistical Software Release 15; StataCorp College Station, TX, USA).P values of <0.05 were considered to indicate significance.

RESULTS
A total of 157,295 men and 183,202 women participants were collected from the 10 cohort studies.The frequency of ever smokers was 79% among men and 13% among women.During 4,729,073 person-years of follow-up, 936 men and 325 women were newly diagnosed with BCa.Men was 3.47 times more likely to develop BCa than women.
Compared with never smokers, former smokers with a duration of smoking cessation of 0-4 years and 5-10 years showed similar HRs to current smokers (HR 1.82; 95% CI, 1.35-2.45and HR 1.81; 95% CI, 1.32-2.50,respectively), as shown in Table 5.On the contrary, former smokers with more than 10 years of smoking cessation had no significantly increased risk compared with never smokers (HR 1.26; 95% CI, 0.97-1.63).In addition, a significant Smoking Intensity, Cessation, and Bladder Cancer in Japan linear trend in decreased BCa risk was observed as years of smoking cessation increased.These findings were not substantially changed when BCa cases diagnosed within the first 3 years after enrolment in each study were excluded.

DISCUSSION
This pooled analysis of 10 cohort studies in Japan showed that both smoking status and cumulative smoking intensity were positively associated with BCa risk.BCa risk increased linearly as cumulative smoking intensity increased.In addition, risk was reduced as the duration of smoking cessation increased, with a significant trend.The risk in former smokers who quit smoking for ≥10 years was comparable with that of never smokers.
A carcinogenic effect of smoking on the bladder is reasonable given that tobacco contains carcinogenic compounds, including   aromatic amines and N-nitroso compounds.These compounds cause DNA damage in the form of double-strand breaks, base modification and bulky adduct formation. 4Our present study epidemiologically confirmed that risk of BCa is increased in association with cigarette smoking.][22][23] On the other hand, the magnitude of the impact of cigarette smoking differed between the Japanese and Western populations.The present HR for current smokers in Japan was similar to those in Asian populations estimated by previous meta-analyses. 5,6The approximately two-fold increase in risk for current smokers in Japan appears lower than the approximately three-fold increase in Western populations.The lower impact of smoking in the Japanese population might be partly explained by differences in the style of inhalation and in age at the initiation of smoking. 11evertheless, the most important determinant of the difference may be variation in carcinogen-detoxification genes between ethnic groups.Analysis of 3,942 cases and 5,680 controls with a European background found six genetic variants with significant additive gene-smoking intearaction, notably in NAT2 and UGT1A6. 24Of the six variants, significant multiplicative genesmoking interaction was observed only in NAT2 (ever vs never smokers, and smoking dose).The NAT2 enzyme plays an important role in detoxifying aromatic amine carcinogens, which are known bladder cancer-related carcinogens in cigarette smoke.The NAT2 genotypes are classified into a fast acetylation type (wild-type), slow acetylation type (mutated-type), and an intermediate acetylation type having both one fast allele and one slow allele.A meta-analysis indicated that the NAT2 slow genotype was associated with increased risk of BCa in Asian as well as Western populations. 25In addition, smokers with the slow NAT2 genotype had higher BCa risk than smokers with the fast/ intermediate NAT2 genotype. 25Of note, the frequency of NAT2 genotypes differed between ethnic groups: the slow genotype accounted for 59% of individuals in Europe but was much less frequent in Northeast Asian countries (China, Japan, and Korea), accounting for 18% of individuals on average. 26This low prevalence of the NAT2 slow genotype in Japan may convincingly explain the lower risk of BCa among smokers compared with Western countries.
Previous epidemoiological studies consistently showed that smoking cessation results in a decrease in BCa risk, consistent with our present findings.][22][23] The recent dose-response meta-analysis by van Osch et al showed that BCa risk substantially decreased as the duration of cessation lengthened, but that former smokers remained at a 50% increased risk compared with never smokers even after 20 years of cessation. 6Our finding that BCa risk for former smokers who quit smoking for ≥10 years was comparable to that of never smokers is inconsistent with the findings of a dose-response metaanalysis that included only one Asian cohort study. 8The difference in the effect of smoking cessation between Western and Japanese populations may be attributable to differences in the prevalence of the NAT2 genotype, as noted above.Reduced exposure to smoking-related carcinogens attributable to the more prevalent fast acetylation enzyme in Japan may allow the BCa risk of former smokers with many years of smoking cessation to approximate that of never smokers.Large-scale cohort studies in other Northeast Asian populations genetically similar to the Japanese population may support our finding.
A strength of our study is its use of pooled analysis using data from large-scale prospective cohort studies.This pooled analysis included approximately 1,000 cases diagnosed with BCa, although previous cohort studies in Asian populations included only several hundred cases.The pooled analysis including a sufficient number of Japanese participants allowed us to robustly estimate BCa risk associated with smoking behaviors using unified categories for cumulative smoking intensity and smoking cessation, although meta-analyses of published articles force the integration of non-uniform populations and categories.Primary prevention by smoking cessation is critical, because no secondary prevention for BCa has yet been established.Given that fewer than half of urology outpatients perceive smoking to be a risk factor for BCa, 27,28 general people may be even more unaware of the BCa risk associated with smoking.Our findings also emphasize the importance of urologist counselling of patients to quit smoking as soon as possible to effectively decrease BCa risk.
Our study has several potential limitations.First, we did not consider potential confounders, such as occupational exposure, a known risk factor for BCa. 4 However, given the low frequency of occupational exposure, the effect of confounding of occupational exposure is unlikely to be remarkable.Second, the participants in the MIYAGI-II study were slightly older generation with shorter duration of follow-up than the other cohort studies.However, the results did not change substantially when the data from the MIYAGI-II study were excluded.Third, incident rates of BCa Smoking Intensity, Cessation, and Bladder Cancer in Japan could have been underestimated because of missing early-stage BCa cases.However, the differences in the underestimation between smokers and non-smokers were unlikely, given bladder cancer is usually detected by hematuria.Thus, if present, this misclassification would have been nondifferential, leading to underestimation of the association with smoking.Fourth, we did not consider competing risks in the present study analyses.Competing risks from death attributable to cardiovascular disease and other cancer may affect the incidence of BCa, given that smoking is associated with several health problems.The beneficial effects of smoking cessation, however, did not alter substantially when we conducted competing risk analysis based on Fine and Gray's model, 29 using data from the major cohorts of the present study, the JPHC-I and JPHC-II studies.Thus, we believe that the risk estimates for smoking cessation in the present study are less biased by competing risks.Fifth, the number of women smokers included in the present study was small, which restricted our analysis to the association between BCa risk and smoking status only, without analysis of cumulative smoking intensity or smoking cessation.Sixth, when additionally adjusted for PY, the similar trend of a dose-response relationship between years since smoking cessation and BCa risk has been observed, but not significant, partly because of insufficient statistical power (data not shown).Future large-scale studies should aim to more closely investigate BCa risk associated with years of smoking cessation, adjusted for potential confounders.
In conclusion, we provide evidence for an association between cumulative smoking intensity and smoking cessation in a doseresponse manner in men using a pooled analysis of 10 population-based cohort studies in Japan.
Data availability: The data underlying this manuscript cannot be shared publicly due to the privacy of study participants.A collaboration with each participating cohort study is required to access the data.

Table 1 .
Characteristics of the cohort studies in the present pooled analysis Aichi Cohort Study; JACC, The Japan Collaborative Cohort Study; JPHC, Japan Public Health Center-based prospective Study; LSS, Life-Span Study; MIYAGI, The Miyagi Cohort Study; OHSAKI, Ohsaki Cohort Study; OSAKA, Osaka Cohort Study; TAKAYAMA, Takayama Study.

Table 2 .
Hazard ratios of bladder cancer risk according to smoking status a Model 1: adjusted for age and public health center area.b Model 2: model 1 excluding early cases (within the first 3 years).+ P < 0.05.

Table 3 .
Hazard ratios of bladder cancer risk according to smoking status and cumulative smoking intensity among current smokers in men a Model 1: adjusted for age and public health center area.b Model 2: model 1 excluding early cases (within the first 3 years).+ P < 0.05.

Table 4 .
Hazard ratios of bladder cancer risk according to cumulative smoking intensity on combination of former and current smokers in men a Model 1: adjusted for age and public health center area.b Model 2: model 1 excluding early cases (within the first 3 years).+ P < 0.05.Masaoka H, et al.J Epidemiol 2023;33(11):582-588 j 585

Table 5 .
Hazard ratios of bladder cancer risk according to years since smoking cessation in men Model 1: adjusted for age and public health center area.b Model 2: model 1 excluding early cases (within the first 3 years).