In this meta-analysis, we found that compared with non-diabetics or general population, individuals with diabetes may have more than 35% increased risk of bladder cancer. However, there were differences in the summary RR among different study designs. Whereas diabetes was positively associated with an increased risk of bladder cancer in case–control and cohort studies, the summary estimate in cohort studies of patients with diabetes did not indicate an excess risk of bladder cancer in these cases compared with the general population.
The strength of the study includes that, on an international scale, there are far more individuals with diabetes in our study. After adjustment for important covariates, our study extends previous meta-analysis by providing a more precise estimate of the association between diabetes and bladder cancer risk (based on 36 studies). Despite similar summary RRs between women and men, the positive association was only observed in men, and was independent of BMI, alcohol consumption, smoking status and physical activity. Lorente and colleagues found that transitional cell carcinoma of the bladder was more frequent in males than females. However, never-smoker women have larger and more aggressive tumors with a higher frequency of muscle-invasive disease than male never-smokers and equaling to male current-smokers . Moreover, women also had higher risk of invasive bladder cancer than men even they smoked comparable amount of cigarettes as men . Further studies are needed to confirm our findings and to understand the molecular pathways that might explain the gender-related differences.
Furthermore, the relationship between duration of exposure to diabetes and risk of bladder cancer could be calculated by summing up data from different studies. The current meta-analysis indicated that the risk of bladder cancer was inversely associated with the duration of DM. There was a slightly increased risk which did not reach statistical significance among individuals with diabetes more than 5 years and an increased rate of bladder cancer was observed in individuals with a shorter duration of diabetes (< 5 years). This finding indicated that those with newly diagnosed diabetes should be highly alert to bladder cancer development. However, MacKenzie and colleagues found that compared with those without diabetes, the risk of bladder cancer was highest among those with diabetes of 16 years or more . It is worth noticing that only three studies in our study presented with RRs for duration of diabetes [10, 24, 33]. When the effect of diabetes was evaluated, glucose-lowering therapies should be adjusted for, and this was not done in most studies. With increasing diabetes duration, the impact of anti-diabetic drugs may set in and influence the association. Thus, the long-term risk of bladder cancer among patients with diabetes warrants further investigation.
In stratified analysis by geographic regions and publication year, we found that the association between DM and bladder cancer was not significant for studies conducted in Europe and for studies published from 1970–1999, whereas studies from North American and Asia and studies published since 2000 showed significantly stronger risk estimates. The regional and temporal differences are perplexing. Many environmental and personal determinants are related, including : genetic factors, lifestyle (eating habits, physical activities, somatotype characteristics), environmental factors (environmental pollution, stress, socioeconomic status ), public health services and so on. With the gradual improvement of medical conditions, early screening and diagnosis rates of DM and bladder cancer are greatly improved. These factors are all attributable to the regional and temporal differences. Our study also has several potential limitations of the available data. Thus, caution is needed when interpreting these results. First, great heterogeneity existed in terms of geographical region, study design, publication year, gender, duration of diabetes and adjustment for confounders. Despite the use of appropriate meta-analytic techniques with random-effect models, we could not account for these differences. The heterogeneity of risk estimates may be due to different mixtures of type 1 and type 2 participants with diabetes and different adjustment for potential confounders. Moreover, some studies included both sexes, whereas others included only men or only women. Nevertheless, subgroup analyses showed that the risk estimate was robust across various quality components.
Second, because diabetes is an underdiagnosed disease, some misclassification of exposure is likely, which would tend to attenuate any true association between diabetes and bladder cancer. In cohort studies of patients with diabetes, the negative association between diabetes and bladder cancer may be due to that the comparison group includes individuals with diabetes, resulting in underestimation of the true effect size.
Third, recent studies have suggested that use of pioglitazone (a common anti-diabetic drug) was associated with an increased incidence of bladder cancer [48, 49]. However, most studies included in this meta-analysis did not adjust for the effect of anti-diabetic drugs, which may distort the true relationship between diabetes and risk of bladder cancer.
Forth, confounding cannot be fully excluded as a potential explanation for the observed association, because our analyses were based on observational studies. It is generally accepted that diabetes and bladder cancer share several common risk factors. Smoking has consistently been associated with increased risk of diabetes and bladder cancer [50, 51]. The relationship between diabetes and bladder cancer was stronger and statistically significant when we restricted the analysis to those studies which controlled for smoking. When risk estimates from the five studies that adjusted for physical activity were combined, the association between diabetes and bladder cancer was also stronger (RR 1.43) than the overall result including all studies (RR 1.35). In the current analysis, however, adjustment for a wide range of potential confounders, including sex, BMI and alcohol consumption, did not significantly alter the relationship between diabetes and risk of bladder cancer.
Fifth, different study designs may have particular methodological issues and constraints; yet, a common theme with all is the potential for bias. Case–control studies are susceptible to recall and selection biases which could inflate the RRs. Cohort studies are prone to be influenced by detection bias because patients with diabetes are under increased medical surveillance. If medical surveillance bias is present, bladder cancer would tend to be diagnosed at an earlier stage in patients with diabetes than in those without diabetes.
Finally, inherent in any meta-analysis of published data is the possibility of publication bias, that is small studies with null results tend not to be published. However, the results obtained from funnel plot analysis and formal statistical tests did not provide evidence for such bias.
Although the absolute risks of bladder cancer are low among individuals with diabetes, our results have important clinical and public health significance. As a serious and growing health problem in USA, DM affects nearly 25.8 million people, 8.3% of the U.S. population in 2010. In China, a cross-sectional study from 2007 through 2008 involving a nationally representative sample of 46,239 adults, the age-standardized prevalences of total diabetes and prediabetes were 9.7 and 15.5%, respectively . Due to growing obesity epidemic, the prevalence of diabetes will probably increase and contribute to the development of bladder cancer.