Impact of diabetes mellitus on recurrence and progression in patients with non-muscle invasive bladder carcinoma: A retrospective cohort study


Seung Il Jung M.D., Department of Urology, Chonnam National University Medical School, 8, Hak-dong, Dong-gu, Gwangju 501-757, Republic of Korea. Email:


Objective:  The aim of the present study was to investigate the relationship between diabetes mellitus (DM) and tumor features in patients with non-muscle invasive bladder cancer (NMIBC).

Methods:  Data from 251 patients who underwent transurethral resection (TUR) for NMIBC from January 2000 to June 2010 were analyzed retrospectively. Patients were divided into two groups: Group I, 159 patients (63%) who did not have DM at the time of surgery; and (ii) Group II, 92 patients (37%) who had DM at the time of surgery. Recurrence- and progression-free survival was assessed in both groups. Preoperative HbA1c levels, as parameter of glycemic control, were determined in Group II patients, with patients divided into two subgroups: (i) HbA1c ≥7.0%; and (ii) HbA1c <7.0%. The clinical features of the bladder tumor were compared in these two subgroups.

Results:  Compared with Group I, Group II patients were older and had a higher rate of hypertension, recurrence, and progression (P < 0.05). Univariate survival analysis showed that gender, DM, smoking, and serum creatinine were associated with recurrence-free survival (P < 0.05), whereas DM, stage, grade, intravesical instillation, and serum creatinine were associated with progression-free survival. In multivariate survival analysis, DM was found to be an independent factor for recurrence- (hazard ratio [HR] 2.11; 95% confidence interval [CI] 1.4–3.2; P = 0.001) and progression-free survival (HR 9.35; 95% CI 3.1–28.6; P = 0.001). Furthermore, patients with HbA1c ≥7.0% exhibited a significantly higher rate of multiplicity (P = 0.001), tumor grade (P = 0.03), and intravesical treatment (P = 0.04).

Conclusions:  In conclusion, DM seems to be an independent predictor of recurrence- and progression-free survival in NMIBC patients. Further prospective studies are needed to establish the prognostic significance of postoperative glycemic control in this patient population.


Approximately 70% of bladder cancers are non-muscle invasive at the time of presentation. Of these, 50%–70% recur within 5 years and 5%–20% progress to muscle invasive disease.1 Therefore, for clinicians who manage non-muscle invasive bladder cancer (NMIBC), individual patient risk assessment for tumor recurrence and progression may guide treatment decisions and surveillance monitoring. Prognostic factors for bladder cancer are tumor size, morphology, grade, multiplicity, stage, with or without carcinoma in situ (CIS), and frequency of recurrence.2

Diabetes mellitus (DM) is a chronic metabolic disease, the incidence of which is increasing. It is thought that DM can affect the treatment outcomes of malignancies. Epidemiological evidence indicates that type 2 DM is a risk factor for several cancers,3–6 but results for bladder cancer have been difficult to interpret. Several studies have proposed an epidemiological association between DM and bladder cancer.7–10 Indeed, if DM and bladder cancer risk have a certain association, it can be hypothesized that DM may have an influence on the recurrence or progression of bladder cancer. In addition, DM-related factors, such as glycemic control, may have significant impact on bladder cancer characteristics. However, there are few data regarding the relationship between DM and the recurrence or progression of bladder cancer.

Therefore, in the present study we investigated the potential association of a history of DM and glycemic control, represented by HbA1c level, with tumor characteristics, including recurrence- and progression-free survival, in patients undergoing transurethral resection (TUR) for NMIBC.


The study was performed in accordance with the Declaration of Helsinki.

Study population

Data of patients with no history of another malignancy who underwent TUR for bladder tumor between January 2000 and June 2010 at Chonnam National University Hospital and Chonnam National University Hwasun Hospital were reviewed retrospectively. Patients were eligible for inclusion in the study if they met the following criteria: (i) histologically diagnosed as urothelial cancer; (ii) complete resection of the disease and first TUR at our institution as the initial treatment for their condition; and (iii) pathologic stage Ta, T1 disease (all T1 specimens had muscle layer). Patients who did not have urothelial cancer, had synchronous upper urinary tract urothelial cancer, or had CIS only and muscle invasive disease were excluded from the study. Using these criteria, 251 patients were enrolled in the present study (Fig. 1).

Figure 1.

Patient enrollment and stratification. TUR, transurethral resection; CIS, carcinoma in situ; DM, diabetes mellitus.

Patients receiving intravesical therapy were subdivided into two groups on the basis of the therapeutic agent administered: (i) Bacillus Calmette–Guérin (BCG), given over a 6-week induction schedule with or without a maintenance schedule; or (ii) epirubicin, given over an 8-week induction schedule with or without a maintenance schedule. Patients who received intravesical treatment underwent follow-up cystoscopy after the induction schedule and every 3 months thereafter.

Data collection

Patients were divided into two groups according to diabetic status: Group I (n = 159), non-DM; and Group II (n = 92), DM. Medication history and preoperative fasting glucose levels were reviewed in Group I patients to confirm that they did not have DM at the time of surgery; Group II patients were identified as having DM at the time of surgery (all type 2 DM) on the basis of a history of DM or medical therapy, such as oral hypoglycemic agents or insulin, and preoperatively elevated fasting glucose levels (>126 mg/dL). Patients who developed DM after initial TUR were excluded from the study. In addition, Group II patients were stratified on the basis of HbA1c levels <7.0% (n = 43) and HbA1c ≥7.0% (n = 49). HbA1c levels were determined at the time of admission for initial TUR for NMIBC. No information was collected regarding the age at diagnosis of DM.

The clinicopathological factors assessed were: (i) age at the time of surgery, gender, history of smoking, comorbidities (DM, hypertension, serum creatinine); (ii) tumor-related factors, including multiplicity, size, pathologic stage (based on the 1997 TNM classification11), grade (based on the 2003 World Health Organization (WHO)/International Society of Urological Pathology Consensus (ISUP) classification12), with or without intravesical therapy after TUR; and (iii) disease status factors, such as recurrence and progression. The association between these various clinicopathological features and preoperative HbA1c levels was also investigated.

Statistical analysis

Data were analyzed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). The distribution of clinicopathological covariates between two groups was evaluated by the Chi-squared test for categorical variables.

Kaplan–Meier analysis was used to assess factors affecting recurrence- and progression-free survival. Recurrence-free survival (RFS) was defined as the time from initial TUR to first tumor recurrence (regardless of grade or stage), whereas progression-free survival (PFS) was defined as the time from initial TUR to tumor progression (any increase in grade [G1/2 to G3] or stage [Ta to T1 or T2, T1 to T2]) after repeat TUR for recurrence. Factors included in the univariate survival analysis were age, gender, smoking, DM, hypertension, intravesical treatment, stage, grade, size, multiplicity, and serum creatinine. Multivariate regression analysis using Cox's proportional hazards regression model (the stepwise forward procedure) was performed to determine the adjusted hazard ratio (HR) to identify prognostic factors for RFS and PFS.

The Mann-Whitney and Chi-squared tests were used for subgroup analysis of patients in Group II (stratified according to HbA1c levels). In all cases, two-tailed P < 0.05 was considered significant.


Baseline characteristics in Groups I and II

The median age of all 251 patients enrolled in the study was 67 years, with a median follow-up time of 34 months (range 1–289 months). Of the 251 patients, 92 (36.7%) were determined to have DM at the time of surgery (Group II) and 159 (63.3%) did not have DM (Group I). The baseline characteristics were similar in both groups in terms of gender, smoking status, serum creatinine, tumor multiplicity, size, stage, grade, and intravesical treatment. However, compared with Group I patients, Group II patients were older, had a shorter follow-up time, and higher rates of hypertension, recurrence, and progression (P < 0.05; Table 1).

Table 1.  Characteristics of patients with bladder tumors
 Total (n = 251)Group I (n = 159)Group II (n = 92)P-value
  • Unless indicated otherwise, data are given as the number of patients in each group, with the percentage given in parentheses.

  • Chi-squared test.

  • Group I, patients without diabetes mellitus (DM); Group II, patients with DM; PUNLMP, papillary urothelial neoplasm of low malignant potential; BCG, Bacillus Calmette–Guérin.

Median age (years)676768 
Median follow-up time (months)343630.5 
Male : female ratio219:32139:2080:120.9
Smokers107 (42.6)71 (44.7)36 (39.1)0.87
Unknown smoking history17 (6.8)5 (3.1)12 (13.0) 
Hypertension90 (35.9)39 (24.5)51 (55.4)0.001
Mean (±SD) serum creatinine (mg/dL)1.1 ± 0.71.0 ± 0.61.2 ± 0.90.1
Tumor multiplicity    
 1–2136 (54.2)87 (54.7)49 (53.3)0.85
 >3113 (45.0)71 (44.7)42 (45.7)
 Unknown2 (0.8)1 (0.6)1 (1.1)
Tumor size    
 <3 cm188 (74.9)117 (73.6)71 (77.2)0.53
 ≥3 cm60 (13.9)40 (25.2)20 (21.7)
 Unknown3 (1.2)2 (1.3)1 (1.1)
pT stage    
 Ta160 (63.7)106 (66.7)54 (58.7)0.2
 T191 (36.3)53 (33.3)38 (41.3)
 PUNLMP14 (5.6)10 (6.3)4 (4.3)0.38
 Low157 (62.5)103 (64.8)54 (58.7)
 High80 (31.9)46 (28.9)34 (37.0)
Intravesical treatment    
 None87 (34.7)60 (37.7)27 (29.3)0.22
 BCG128 (51.0)80 (50.3)48 (52.2)
 Epirubicin36 (14.3)19 (11.9)17 (18.5)
Recurrence93 (39.0)51 (32.1)47 (51.1)0.003
Progression22 (8.8)4 (2.5)18 (19.6)0.001

Association between clinical parameters and RFS and PFS

Univariate survival analysis showed that male gender, DM, smoking, and serum creatinine (>1.5 mg/dL) were associated with shorter RFS (P < 0.05, log rank test; Table 2) and that DM, pT1, high grade, no intravesical instillation, and serum creatinine (>1.5 mg/dL) were associated with shorter PFS (P < 0.05, log rank test; Table 2).

Table 2.  Factors affecting recurrence- and progression-free survival identified using univariate survival analysis
FactorsRecurrence-free survival (months)Progression-free survival (months)
Mean (95% CI)P-valueMean (95% CI)P-value
  1. DM, diabetes mellitus; PUNLMP, papillary urothelial neoplasm of low malignant potential; BCG, Bacillus Calmette–Guérin.

 <Median age65.6 (56.2–75.0)0.09113.5 (101.2–125.9)0.26
 ≥Median age52.2 (44.3–60.1)246.2 (221.4–271.0)
 Male54.5 (47.5–61.5)0.04237.9 (202.1–272.4)0.8
 Female75.3 (61.0–89.5)90.0 (80.4–99.6)
 Yes41.9 (35.3–48.5)0.03273.9 (261.0–286.8)0.21
 No63.1 (54.4–71.7)168.9 (145–192.8)
 Positive40.4 (59.1–73.8)0.00187.5 (69.8–105.3)0.001
 Negative66.5 (59.1–73.8)281.0 (273.1–288.8)
 Yes43.2(34.9–51.6)0.0572.8 (67.4–78.1)0.11
 No62.0 (54.5–69.5)246.6 (217.7–275.4)
Intravesical treatment    
 BCG57.0 (48.1–65.8)0.06161.8 (132.3–191.3)0.025
 Epirubicin36.5 (26.1–46.8)270.5 (245.8–295.2)
 None66.3 (56.3–76.3)119.6 (109.1–130.2)
pT stage    
 Ta61.5 (53.8–69.1)0.26275.8 (265.3–286.2)0.001
 T152.4 (42.2–62.6)143.8 (108.4–179.2)
 PUNLMP–low61.5 (54.1–68.9)0.15272.5 (261.3–283.8)0.001
 High43.3 (35.7–50.8)102.7 (52.4–153.0)
 <3 cm57.9 (50.6–65.2)0.56245.9 (224.0–267.7)0.08
 ≥3 cm58.0 (47.7–68.2)86.2 (82.6–89.9)
 <364.9 (56.6–73.1)0.051259.5 (236.3–282.7)0.28
 ≥348.6 (40.4–56.7)162.4 (149.3–175.4)
 ≤1.5 mg/dL59.8 (53.4–66.3)0.007246.8 (220.8–273.6)0.049
 >1.5 mg/dL21.4 (10.9–31.9)59.0 (43.7–74.4)

Multivariate survival analysis (Table 3) revealed that smoking (P = 0.023) and DM (P = 0.001; Fig. 2) were associated with an increased risk of recurrence, whereas pT1 stage (P = 0.003) and DM (P = 0.001; Fig. 2) were associated with an increased risk of progression.

Table 3.  Recurrence- and progression-free survival in patients with non-muscle invasive bladder cancer (multivariate survival analysis)
FactorsHazard ratio (95% CI)P-value
  1. DM, diabetes mellitus.

Recurrence-free survival  
 Smoking1.63 (1.1–2.5)0.023
 DM2.11 (1.4–3.2)0.001
Progression-free survival  
 pT stage (T1)4.70 (1.6–13.2)0.003
 DM9.35 (3.1–28.6)0.001
Figure 2.

Relationship between diabetes mellitus (DM) and survival in patients with non-muscle invasive bladder cancer. The blue lines indicate patients without DM, whereas the green lines indicate patients with DM.

Subgroup analysis of DM patients according to HbA1c levels

Differences in the clinical variables were evaluated between the two DM subgroups, stratified according to HbA1c levels before initial TUR (≥7.0% vs <7.0%); 43 patients were found to have good glycemic control (HbA1c <7.0%) and 49 were found to have poor glycemic control (HbA1c ≥7.0%). Compared with patients who had good glycemic control, those with poor glycemic control exhibited a significantly higher rate of multiplicity (P = 0.001), tumor grade (P = 0.03) and intravesical treatment (P = 0.04; Table 4).

Table 4.  Characteristics of subjects with both diabetes mellitus and bladder cancer stratified according to preoperative HbA1c levels
 HbA1c <7.0% (n = 43)HbA1c ≥7.0% (n = 49)P-value
  • Unless indicated otherwise, data are given as the number of patients in each group, with the percentage given in parentheses.

  • *

    Mann-Whitney test.

  • Chi-squared test.

  • PUNLMP, papillary urothelial neoplasm of low malignant potential.

Stage (Ta/T1)28 (65.1)/15 (34.9)26 (53.1)/23 (46.9)0.24
Grade (PUNLMP/low/high)3 (7.0)/30 (69.8)/10 (23.3)1 (2.0)/24 (49.0)/24 (49.0)0.03
Size (<3 cm/≥3 cm/unknown)32 (74.4)/11 (25.6)39 (79.6)/9 (18.4)/1 (2)0.43
Multiplicity (<3/≥3/unknown)31 (72.1)/12 (27.9)18 (36.7)/30 (61.2)/1 (2)0.001
Intravesical treatment26 (60.5)39 (79.6)0.04
Recurrence18 (41.9)29 (59.2)0.09
Mean (±SD) interval until first recurrence (months)23.1 ± 23.517.1 ± 14.50.15*
Early recurrence (within 1 year)13 (30.2)15 (30.6)0.96
Mean (±SD) progression interval (months)32.8 ± 27.527.6 ± 22.10.31*
Progression6 (14.0)12 (24.5)0.2


The results of the present study show that DM is an independent prognostic factor for RFS and PFS. In addition, patients with poor glycemic control (HbA1c ≥7%) had more high-grade tumors, greater tumor multiplicity, and, consequently, received more intravesical chemotherapy than patients with good glycemic control (HbA1c <7%). The prevalence of diabetes mellitus and cancer is increasing in current society. It is not surprising that the incidence of the concomitant occurrence of two diseases will rise. Diabetes was found to be common in some cancers, including breast, colorectum, endometrium, liver, and pancreatic cancers.3–6

Epidemiologic studies of diabetes and the risk of bladder cancer have reported inconsistent results. Several cohort studies have reported that diabetes is related to a significant 1.3–2.5-fold increased risk of bladder cancer.13–15 Conversely, others have failed to find any association between the two disease entities.10,16,17 In addition, case control studies relating diabetes to bladder cancer risk have yielded inconsistent results, with a significant positive relationship found in one study,7 but opposite results reported in another.18

The present study focused on DM and whether it affected RFS and PFS. The risks for both recurrence and progression in NMIBC are associated with multiple tumor-related factors, including histologic grade, depth of invasion, multiplicity, tumor size, tumor morphology, the presence or absence of vascular or lymphatic invasion, and the presence or absence of CIS. Although these tumor-related factors provide some prognostic information, they ultimately fail to clearly evaluate the malignant potential of individual tumors. Therefore, there is a need for other clinical factors to predicting tumor recurrence or progression in NMIBC. Chang et al. examined the relationship between smoking and tumor features and the risk of recurrence among men with bladder cancer.19 They found that tumor features were not associated with smoking, but RFS was shorter in current smokers compared with non-smokers and ex-smokers. Migaldi et al. reported that young patients (<45 years) with NMIBC had a better prognosis than those >55 years of age, exhibiting a lower grade and stage at diagnosis and a lower recurrence rate.20 In their study of clinical risk factors, Joo et al. reported that patients with NMIBC who had high serum creatinine levels (>1.5 mg/dL) had more frequent recurrence, but not progression, than those with low serum creatinine levels.21 In addition, those authors suggested that DM may affect tumor recurrence (hazard ratio [HR] 2.06; 95% confidence interval [CI] 0.89–4.72; P = 0.091). Our results for DM reinforce these findings (for recurrence, HR 2.11, 95% CI 1.4–3.2, P = 0.001; for progression, HR 9.35, 95% CI 3.1–28.6, P = 0.001), our findings regarding serum creatinine are inconsistent with those of Joo et al., with no significant association found between serum creatinine and recurrence or progression in the present study. To clarify this issue may require a large prospective cohort study. Another important finding of the present study was that there were no differences in RFS according to intravesical treatment type, but that in the case of PFS, BCG-treated patients had lower PFS than epirubicin-treated patients. This may be explained by a greater incidence of pT1 and high-grade disease in the BCG treated group than in the epirubicin-treated group.

An additional purpose of the present study was to evaluate tumor characteristics according to HbA1c, which reflects glycemic control. Poor glycemic control was associated with tumor-related recurrence or progression risk factors (e.g. grade, multiplicity) and, as a result, more patients with poor glycemic control received intravesical treatment than patients with good glycemic control. However, we did not seek to assess clinical variables according to changes in HgA1c level. This should be determined through future prospective studies.

The importance of glycemic control has been highlighted in other types of cancer. Wu et al. reported that DM is a prognostic factor for oral squamous cell carcinoma from both aspects of survival and recurrence, but they suggested that HbA1c levels could be a more reliable indicator.22 Siddiqui et al. reported that poor glycemic control, as judged by the HbA1c level, independently predicted the early onset of colorectal cancer, a more advanced stage at the time of presentation, and poorer 5-year survival.23 In addition, Hong et al. compared pre-radical prostatectomy HbA1c levels with several measures of disease aggressiveness and observed that elevated HbA1c predicted higher pathological Gleason score and extraprostatic tumor extension in prostate cancer;24 their results for higher pathological Gleason scores were verified by Kim et al.25

The mechanism by which DM contributes towards bladder cancer remains unknown. One suggestion is that chronic exposure to hyperinsulinemia or hyperglycemia induces tumor cell proliferation and metastasis,26 and that increased insulin-like growth factor (IGF)-1 in diabetic patients stimulates cellular proliferation and inhibits apoptosis.27 Administration of IGF-1 resulted in a shortened latency period and more rapid growth of the tumors in male NIH athymic nude mice; these effects were particularly noticeable in tumors resulting from cells that expressed higher levels of IGF-1 receptors.28 Another possible mechanism involves structural changes in the urothelium in diabetic nephropathy. Cadherins are a family of membrane glycoproteins involved in cell–cell adhesion. Singh et al. looked at cadherin distribution and quantity in rat renal cells exposed to glycated proteins and found a decrease in both the amount and distribution of cadherins.29 In addition, some studies have revealed that degraded glycosaminoglycans may promote bladder cancer progression and metastasis.30

If cadherin expression and degradation of glycosaminoglycans in bladder tissue is affected by DM, this could explain the poor prognosis of patient with DM. Reduced expression of E-cadherin is associated with poor outcomes in bladder cancer patients and has been shown to be correlated with increased tumor invasion.31,32 Furthermore, it was proposed that metabolic factors, such as glucose and IGF-1 led to decreased production of the celluar component of basement membrane and increased degradation.7 This mechanism may also be relevant to patients with DM and NMIBC. However, further investigations are warranted regarding structural changes to the bladder urothelium in patients with DM and NMIBC.

Diabetes is also associated with an increased risk of urinary tract infections (UTI),33 which has been related to bladder cancer risk,7 and UTI may affect the incidence of recurrence or progression. However, we could not analyze these factors in the present study because of missing data (UTI history, urine culture results).

The present study is not without limitations. First, in addition to its retrospective nature, there was a relatively small number of patients enrolled in the study, so we did not analyze other factors that may have been meaningful, including lymphovascular invasion, mixed histologic variants (and percentage), aspirin use, statin use, restaging transurethral resection of bladder tumor for T1, and high-grade Ta (bulky) disease. In addition, although previously reported tumor-related risk factors (except smoking and pT stage) were not associated with RFS or PFS in the present study, subgroup analysis showed trends for these factors. Second, the frequency of tumor recurrence and progression was low in the patients in the present study, so we used mean survival time instead of median survival time. Third, we did not assess the duration of DM, which may have been affected by recall bias or a period of undetected DM. Fourth, our results have a selection bias because of possibilities of undiagnosed DM among patients classified as not having DM (Group I). However, to our knowledge, there is limited literature evaluating the association between DM and RFS and PFS in NMIBC; thus, the present small study may provide important information to clinicians treating NMIBC patients with DM.


The present study has shown that DM is an independent predictor of RFS and PFS for NMIBC. Therefore, close follow-up and intravesical therapy may be beneficial in patients with DM, especially those with poor glycemic control. In addition, large prospective studies are needed to elucidate the effect of DM and glycemic control on the recurrence and progression of NMIBC.


This study was supported by a grant (CRI10064-1) from the Chonnam National University Hospital Research Institute of Clinical Medicine.

Conflict of interest

None declared.