SEARCH

SEARCH BY CITATION

Keywords:

  • bladder cancer;
  • cohort;
  • type 2 diabetes;
  • epidemiology

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Type 2 diabetes mellitus (T2DM) is associated with increased bladder cancer incidence in some, but not all, studies. Many studies had limited statistical power and few examined risk by insulin-use, duration of diabetes or cancer stage. We examined the association between T2DM and bladder cancer incidence in the Cancer Prevention Study II Nutrition Cohort, a large prospective study with information on insulin-use and duration of diabetes. Diabetes and insulin-use were ascertained from a questionnaire at study enrollment in 1992 or 1993 and updated in 1997 and every 2 years thereafter. During follow-up through 2007, 1,852 cases of incident bladder cancer were identified among 172,791 participants. Multivariable adjusted relative risks (RRs) and 95% confidence intervals (CIs) were estimated using extended Cox regression modeling. There were no associations of T2DM with the risk of bladder cancer overall (RR = 1.01, 95% CI: 0.87–1.17), noninvasive disease (RR = 0.93, 95% CI: 0.76–1.14) or invasive disease (RR = 1.13, 95% CI: 0.91–1.40). Compared to participants without T2DM, risk of invasive bladder cancer was higher among participants who had had T2DM for >15 years (RR = 1.63, 95% CI: 1.09–2.43) and among those using insulin (RR = 1.64, 95% CI: 1.18–2.27). These results do not support an association of T2DM with overall bladder cancer incidence, but do suggest positive associations of long-term T2DM and insulin-use or other factors correlated with severe diabetes, with invasive bladder cancer incidence.

Type 2 diabetes mellitus (T2DM) has been associated with higher risk of cancer at several organ sites, including the liver, pancreas, endometrium and colorectum,1 and to a lesser extent, bladder cancer. A 2006 meta-analysis found that T2DM was associated with a statistically significant 24% higher risk of bladder cancer compared to no T2DM.2 Since the meta-analysis was published, four new studies reported statistically significantly higher risks3–6 while three other studies reported no association.7–9

Although the biological mechanisms underlying an association between T2DM and bladder cancer are unclear, it has been suggested that high circulating levels of insulin and/or insulin-like growth factor-1 might play a role.2, 10 Insulin levels, however, are increased in obese individuals,11 and obesity has not been generally associated with the risk of bladder cancer.12 Alternatively, men and women with T2DM develop urinary tract infections more frequently,13 which could in turn result in higher risk of bladder cancer.14

Relatively few studies have been able to examine the association between T2DM and bladder cancer by diabetes duration4, 6, 9 or insulin-use,4, 6, 15 and results from these studies have been mixed. Some studies suggest that longer duration diabetes6 and insulin-use4 are associated with higher risk, while others report no difference by diabetes duration4 or insulin-use.6, 15 To our knowledge, no study has examined duration of diabetes or insulin-use, which might play a role in progression of cancer, by bladder cancer stage.

We examined the association of T2DM, duration of T2DM and insulin-use with the risk of bladder cancer overall and by stage of disease among men and women in the Cancer Prevention Study II (CPS-II) Nutrition Cohort. The CPS-II Nutrition Cohort is well-suited to examine these associations due to its large size and the availability of regularly updated information on diabetes diagnosis and insulin-use.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Study cohort

Men and women for this analysis were drawn from ∼184,000 participants in the CPS-II Nutrition Cohort, a prospective study of cancer incidence in the United States established in 1992 and described in detail elsewhere.16 The Nutrition Cohort is a subgroup of the ∼1.2 million participants in the CPS-II, a prospective study of cancer mortality established in 1982 by the American Cancer Society. The CPS-II Nutrition Cohort has been approved by the Emory University Institutional Review Board.

At enrollment in 1992 or 1993, CPS-II Nutrition Cohort participants completed a mailed self-administered questionnaire that included information on demographic, medical and lifestyle factors. Follow-up questionnaires were sent to cohort members in 1997 and every 2 years thereafter to update exposure information and to ascertain newly diagnosed cancers. Response rates among participants who were sent follow-up questionnaires (all living participants who had not been lost to follow-up or requested permanent withdrawal from study follow-up) were at least 87% for each follow-up questionnaire.

This analysis excluded participants with a history of bladder cancer at enrollment (n = 924), those who self-reported bladder cancer on the first follow-up survey and could not be verified through medical record abstraction or cancer registry linkage (n = 71), those who were lost to follow-up (alive at the time of the first follow-up questionnaire in 1997, but did not complete the 1997 or any subsequent follow-up questionnaire, n = 6,240) and those with missing information on year of bladder cancer diagnosis (n = 2). In addition, we excluded those with inconsistent diabetes data (reported diabetes in 1982 but not in 1992, n = 1,009) or missing diabetes information at enrollment (n = 637) and those with unknown smoking status at baseline (n = 2,304). We also excluded participants who reported insulin-use before age 30 years (n = 210) to reduce potential bias from including patients that likely had type I diabetes. The final analytic cohort included 80,805 men and 91,986 women.

Ascertainment of cancer cases

This analysis included 1,852 incident cases of bladder cancer (1,502 men, 350 women) diagnosed between the date of enrollment and June 30, 2007. Of these, 1,656 cases were initially identified by self-report on a follow-up questionnaire and subsequently verified by obtaining medical records or through linkage with state cancer registries when complete medical records could not be obtained.16 Nutrition Cohort participants have been shown to report cancer diagnoses with high sensitivity (0.93).17 An additional 196 cases of cancer were identified through linkage with the National Death Index,18 of which 139 were verified through subsequent linkage with state cancer registries. Bladder cancers were classified as invasive if they were stages I, II, III or IV at diagnosis according to the American Joint Committee on Cancer (AJCC)19 (n = 761), or if bladder cancer was listed as the underlying cause of death on the death certificate, and no information on stage was available from medical records or record linkage (n = 51) or as noninvasive if they were AJCC stage 0 (Tis or Ta) at diagnosis (n = 992). Cancers with insufficient information on stage (N = 48) were censored from stage-specific analyses.

Assessment of DM and insulin

Diabetes status was reported on questionnaires in 1982 (at the time of enrollment into the larger CPS-II mortality cohort), 1992–1993 (baseline-at enrollment in the Nutrition Cohort), 1997, 1999, 2001, 2003 and 2005. The 1982 and 1992–1993 questionnaires asked if participants had ever been diagnosed with diabetes by a physician. The 1997 questionnaire also asked about timing of diabetes diagnosis using response categories of “before 1992,” “1992–1993,” “1994–1995” or “1996 or after.” On the 1997 questionnaire and on all subsequent surveys, the question was modified to exclude gestational diabetes. Participants were asked if they had used insulin beginning with the baseline questionnaire (1992–1993) and on all subsequent surveys. Participants were not asked to specify the type or dose of insulin or about other drugs used to lower or maintain glucose levels.

Statistical analysis

Age-adjusted and multivariable-adjusted RRs and 95% confidence intervals (CIs) were calculated using extended Cox regression models to estimate the associations of T2DM, duration of T2DM (time since diagnosis) and insulin-use with the risk of bladder cancer. Follow-up time for Cox models began on the date of completion of the 1992–1993 questionnaire.

T2DM diagnosis, duration of T2DM and insulin-use were modeled using time-dependent variables initially defined in 1992–1993 and updated in 1997, 1999, 2001, 2003 and 2005. Participants without T2DM served as the main referent group for analyses. To calculate T2DM duration, participants who first reported diabetes on the 1982 questionnaire were initially assigned 10 years since diagnosis, while those who first reported diabetes on the 1992 questionnaire were assigned 5 years since diagnosis. Participants who first reported diabetes on the 1997 questionnaire were assigned a value for years since diabetes diagnosis based on the diagnosis year reported on that questionnaire. Those who first reported diabetes on the 1999 or subsequent questionnaires were assigned a diabetes diagnosis date midway between the date of the questionnaire on which they reported diabetes and the date of the previous questionnaire. Once a participant indicated a diabetes diagnosis, time since diagnosis was extended until the participant was diagnosed with bladder cancer or was censored from the analysis. Time since diagnosis was classified as ≤5 years, >5–10 years, >10–15 years and >15 years. Likelihood ratio tests were used to assess linear trends with increasing duration of diabetes, excluding those without T2DM. In models that examined insulin-use, participants with T2DM were categorized into two groups: those who reported ever using insulin and those who did not.

All multivariable analyses were adjusted for sex, race (white, black or other), BMI (<18.5, 18.5–<25, 25–<30, 30–<35, ≥35 kg/m2 or unknown), educational level (less than high school graduate, high school graduate, some college, college graduate, graduate school or unknown) alcohol use (nondrinker, <1 drink per day, 1 drink per day, >1 drink per day or unknown) and smoking status {never smoker, former smoker (subcategorized by years since quit: ≤5, >5–10, >10–15, >15–20, >20–25, >25 or unknown) or current smoker [subcategorized by combinations of years smoked (=35 or >35) and cigarettes per day (<20 or =20)]{. The stratified Cox procedure20 was used to adjust for age with 1-year strata. All covariates were modeled based on baseline values. In additional analyses, inclusion of physical activity, use of nonsteroidal anti-inflammatory drugs, multivitamin use, fruit and vegetable intake, red meat intake and geographic region had negligible effects on the relative risk (RR) estimates and were not included in the final multivariable models.

We examined whether the association between T2DM and bladder cancer risk varied by sex, smoking status (never, current or former) or follow-up time (continuous). Specifically, we modeled multiplicative interaction terms between T2DM, and each of these factors and calculated a p value for interaction by comparing the likelihood ratio statistic from models with and without interaction terms. All p values are two-sided; p values less than 0.05 are considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Select descriptive data for participants with and without T2DM at baseline are shown in Table 1. At baseline, 7.4% of participants reported physician-diagnosed diabetes. Compared to those without T2DM, participants with the disease were on average older, less educated, less likely to drink alcohol, less physically active and more likely to be male, obese or an ever smoker. Participants were followed for a mean of 11.9 years (median = 13.8 years).

Table 1. Age-adjusted percentages1 of baseline characteristics by type 2 diabetes mellitus (T2DM) status at baseline in the Cancer Prevention Study II Nutrition Cohort, 1992–2007
inline image

There were no associations of T2DM with risk of overall, noninvasive (stage 0) and invasive (stages I–IV) bladder cancers (Table 2). In analysis restricted to stages II–IV bladder cancer (236 cases), the association remained null (RR = 1.20, 95% CI: 0.81–1.77). The association between T2DM and overall bladder cancer was similar when adjusted only for age and sex (RR = 1.03, 95% CI: 0.90–1.19) compared to multivariable adjusted analysis. Associations between T2DM and overall bladder cancer did not differ between men (RR = 0.99, 95% CI: 0.85–1.16) and women (RR = 1.15, 95% CI: 0.80–1.67) and were similar when stratified by smoking status (p-heterogeneity = 0.99). In analyses stratified by follow-up time, the RR for overall bladder cancer was 1.12 (95% CI: 0.89–1.41) during the first half of follow-up (1992–1999) and 0.95 (95% CI: 0.79–1.15) during the second half of follow-up (2000–2007). For invasive bladder cancer, there was some suggestion of higher risk from 1992 to 1999 (RR = 1.36, 95% CI: 0.99–1.87), but not from 2000 to 2007 (RR = 0.98, 95% CI: 0.74–1.31), although a test for interaction by follow-up time (modeled as a continuous variable) was not statistically significant (p = 0.29).

Table 2. Bladder cancer incidence by type 2 diabetes mellitus (T2DM) and insulin use in the Cancer Prevention Study II Nutrition Cohort, 1992–20071
inline image

Associations with overall and noninvasive bladder cancer did not differ by duration of T2DM or by insulin-use (Table 2). Participants who had diabetes for more than 15 years were at higher risk of invasive bladder cancer compared to those without T2DM (RR = 1.63, 95% CI: 1.09–2.43). Similarly, participants with T2DM who used insulin were at higher risk of invasive bladder cancer compared to those without T2DM (RR = 1.64, 95% CI: 1.18–2.27). In analyses restricted to those with T2DM, insulin-use remained statistically significantly associated with higher risk of invasive bladder cancer (RR = 1.70, 95% CI: 1.13–2.54). Associations with both duration of T2DM and insulin-use appeared to be slightly stronger when restricted to stages II–IV bladder cancer (for duration of T2DM > 15 years, RR = 1.98, 95% CI: 1.03–3.80; for insulin-use, RR = 2.38, 95% CI: 1.41–4.00).

To better understand the independent associations of duration of T2DM and insulin-use, we examined risk of invasive bladder cancer by combination of duration of diabetes and insulin-use (Table 3). For those without insulin-use, risk of invasive bladder cancer was higher in those with long duration T2DM compared to those without T2DM, but this association was not statistically significant. Regardless of duration of diabetes, T2DM with insulin-use was associated with a statistically significantly higher risk of invasive bladder cancer.

Table 3. Incidence of invasive bladder cancer1 by duration of type 2 diabetes mellitus (T2DM) and insulin use in the Cancer Prevention Study II Nutrition Cohort, 1992–20072
inline image

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In this large prospective study, there was no association between T2DM and risk of overall or invasive bladder cancer. Long-duration T2DM and insulin-use, however, were associated with statistically significantly higher risks of invasive bladder cancer compared to no T2DM.

Results of studies examining the association between T2DM and overall risk of bladder cancer have been inconsistent. Record linkage studies, none of which had information on smoking, have been mostly null,9, 21–23 with two exceptions.4, 5 In contrast, the majority of case–control and cohort studies support an association between T2DM and higher risk of bladder cancer. Case–control studies6, 15, 24–29 have generally reported diabetes to be associated with higher risk,6, 15, 24–27 with odds ratios ranging from 1.1 to 2.7, although results from some studies were not statistically significant.24, 25 All cohort studies have reported RRs above one,2, 3, 7, 30, 31 generally in the range of 1.2–1.5, with the majority of the studies reporting statistically significant associations.3, 30, 31

It is unclear why there was no association between T2DM and overall risk of bladder cancer in our study. One potential explanation for this finding is that glucose levels among people with diabetes may have been better controlled in our study population than in other study populations. Most participants in our study are relatively well-educated and were eligible for Medicare coverage during the study follow-up period. In addition, the timing of our study follow-up coincided with an apparent improvement in glucose control among patients with diabetes in the United States as measured by hemoglobin A1c.32 Indeed, in our cohort, there was some suggestion of higher risk of bladder cancer in the first half of follow-up (1992–1999), particularly with invasive bladder cancer (RR = 1.36, 95% CI: 0.99–1.87), while there was no suggestion of higher risk from 2000 to 2007.

In our study, long-term T2DM and insulin-use were associated with statistically significantly higher risks of invasive bladder cancer, particularly for more advanced tumours (stages II–IV cancers). Few other studies examined the association of T2DM with bladder cancer by duration of diabetes4, 6, 9 or by insulin-use,4, 6, 15 and no previous study has examined duration or insulin-use by stage of bladder cancer. In a US case–control study that included 331 cases, risk of bladder cancer appeared higher among individuals with diabetes for ≥16 years than among those diagnosed with diabetes more recently, while no apparent difference was observed between insulin users and noninsulin users.6 In a Taiwanese record linkage study that included 589 cases, the association between diabetes and bladder cancer risk did not differ by duration of diabetes, but the highest duration category examined was only ≥5 years.4 The Taiwanese study suggested higher risk of bladder cancer among insulin users compared to people with diabetes who did not use insulin, but this difference was not statistically significant.4 A Canadian case–control study that included 835 cases found similarly increased risk among people with diabetes using insulin and those who did not.15 A US record linkage study found slightly lower risk in those with diabetes for >10 years compared to those without diabetes; however, this study could not adjust for smoking.9

Although both duration of T2DM and insulin-use was associated with higher risk of invasive bladder cancer in this analysis, our results provide stronger evidence for an association with insulin-use, independent of duration, than for an association with duration of diabetes, independent of insulin-use. These findings, however, do not necessarily indicate that insulin itself increases risk. One possibility is that a correlate of insulin-use, such as a history of poor glucose control, contributes to bladder cancer progression. A recent study of patients surgically treated for nonmuscle invasive bladder cancer provides limited support for a role of poor glucose control in bladder cancer progression.33 In that study, patients with diabetes and poor glucose control were significantly more likely than patients with diabetes and good glucose control to have high grade tumors and a higher rate of multiplicity at diagnosis. In addition, patients with diabetes and poor glucose control also appeared more likely than those with diabetes and good glucose control to experience tumor recurrence during the study follow-up period (p = 0.09).33 However, further research will be needed to clarify if insulin and glucose control play a role in bladder cancer progression.

Important strengths of this study include the prospective design and large sample size. To our knowledge, this is the largest prospective study of diabetes and bladder cancer to date that was able to adjust for smoking, a strong risk factor for bladder cancer.12 Additionally, information on diabetes status and insulin-use was regularly updated in this study, reducing the potential for misclassification. An unusual strength of this study was the ability to examine associations by duration of diabetes and insulin-use. Limitations of this study include that diabetes and insulin-use were self-reported, potentially resulting in some misclassification; however, self-reported diabetes agreed well with medical records in a subsample of participants from this cohort.34 In addition, no information was available on use of oral diabetes medications or on markers of diabetes severity such as hemoglobin A1c. Therefore, we were unable to adjust for use of oral medications to examine results by severity of diabetes or to determine if insulin-use was associated with risk after adjustment for severity of diabetes.

In conclusion, our results do not support an association between T2DM and risk of overall bladder cancer, but do suggest that long-term diabetes and insulin-use are associated with higher risk of invasive bladder cancer. Associations with insulin-use should be interpreted with caution as insulin-use is associated with diabetes severity, and we were unable to investigate other markers of diabetes severity such as degree of blood glucose control. Future investigations of T2DM and risk of bladder cancer should carefully examine the results by insulin-use and other markers of diabetes severity, duration of diabetes and bladder cancer stage.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The authors of this work express our sincere appreciation to all CPS-II study participants and to each member of the study management group. The American Cancer Society funds the creation, maintenance and updating of the CPS-II.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References