Secondary bladder cancer after upper tract urothelial carcinoma in the US population

Authors


James M. McKiernan, Columbia University, Department of Urology, 161 Fort Washington Avenue, Herbert Irving Pavilion 11th Floor, New York, NY 10032, USA. e-mail: jmm23@columbia.edu

Abstract

Study Type – Prognosis (cohort)

Level of Evidence 2a

What's known on the subject? and What does the study add?

It is known that a certain percentage of patients treated for upper tract urothelial carcinoma (UTUC) will go on to develop a secondary bladder cancer; however, the risk factors for developing a secondary bladder tumour have not been studied in a population-based setting.

Given the large changes in how UTUC has been diagnosed and managed in recent years, this study aimed to evaluate the natural history of UTUC in the US population over a 30-year period, with a particular emphasis on the development of secondary bladder cancer.

OBJECTIVE

  • • To assess the natural history of upper tract urothelial carcinoma (UTUC) and the development of lower tract secondary cancer.

PATIENTS AND METHODS

  • • Patients diagnosed with UTUC between 1975 and 2005 were identified within nine Surveillance, Epidemiology and End Results registries.
  • • Baseline characteristics of patients with and without secondary bladder cancer were compared.
  • • A multivariate logistic regression model was fitted to test if the year of diagnosis predicted the likelihood of developing a secondary bladder cancer.

RESULTS

  • • Of the 5212 patients with UTUC, 242 (4.6%) had a secondary bladder cancer (range: 1.7–8.2%).
  • • There was a mean interval of 26.5 (95% CI: 22.2–30.8) months between cancer diagnoses.
  • • Compared with those without secondary tumours, patients with secondary bladder malignancy were more likely to present with larger tumours (4.2 vs 3.1 cm, P < 0.001) and with tumours located in the ureter (P < 0.001).
  • • Year of diagnosis was not a predictor of the likelihood of having a secondary bladder malignancy in a multivariate analysis controlling for demographic and tumour characteristics (odds ratio: 0.99; 95% CI: 0.95–1.03)

CONCLUSIONS

  • • Patients with larger urothelial tumours located in the ureter were those most likely to develop a secondary lower tract tumour.
  • • No longitudinal changes in the rate of secondary bladder cancer were noted among patients with UTUC over the 30-year study period.
Abbreviations
UTUC

upper tract urothelial carcinoma

SEER

Surveillance, Epidemiology and End Results

OR

odds ratio

INTRODUCTION

Upper tract urothelial carcinoma (UTUC), is an aggressive, relatively rare neoplasm that comprises <5% of all urothelial cancers [1]. Over the past 30 years, there have been dramatic changes in both the diagnosis and management of this cancer. Many of the diagnostic techniques used today, including the now widely used ureteroscopy and advanced cross-sectional abdominal imaging, were not widely available as recently as the early 1980s [2,3]. Similarly, treatment strategies for UTUC have undergone significant development with the advent of laparoscopic nephro-ureterectomy as well as primary endoscopic treatments for low grade disease [4,5]. During this era of technological advancement, cigarette smoking, which constitutes the primary risk factor for developing UTUC, has been on the decline in the USA.

It is known that a certain percentage of patients treated for UTUC will go on to develop a secondary bladder cancer [6]; however, the risk factors for developing a secondary bladder tumour have not been studied in a population-based setting. Given the large changes in how UTUC has been diagnosed and managed in recent years, we sought to evaluate the natural history of UTUC in the US population over a 30-year period, with a particular emphasis on the development of secondary bladder cancer.

MATERIALS AND METHODS

PATIENT POPULATION

A total of 5212 patients diagnosed with transitional cell carcinoma of the ureter, renal pelvis and kidney were identified within the nine Surveillance, Epidemiology, and End Results (SEER) registries during the study period of 1975–2005. The nine registries were selected based on their presence during the entire time period of interest. The registries included the following geographic areas: Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco/Oakland, Seattle-Puget Sound, and Utah. The characteristics of the SEER population were similar to the US population in general [7]. Patients were excluded if they met any of the following criteria: (i) had received perioperative radiotherapy; (ii) had a history of bladder cancer; (iii) had incomplete follow-up information; (iv) had non-urothelial cell primary histology; or (v) had a known history of metastatic cancer.

STATISTICAL ANALYSIS

A diagnosis of secondary bladder malignancy was determined based on duplicate patient-identifiers in SEER, such that if a patient with UTUC was later diagnosed with bladder cancer in a SEER catchment area, they would have two entries under the same patient identification key. Differences in baseline clinical and pathological characteristics between patients with and without secondary bladder malignancies were assessed using chi-squared analysis. Those without a bladder cancer diagnosis up to 31 December 2005 were classified as censored (no secondary malignancy at the end of follow-up). A multivariate logistic regression model was then created to determine if the year of diagnosis predicted the likelihood of developing a secondary bladder cancer. Statistical analyses were performed using the SAS software program, version 9.1. (SAS Institute Inc, Cary, NC, USA).

RESULTS

A total of 5212 patients with UTUC were identified within the nine SEER registries. From this aggregate, 242 (4.6%) had a secondary bladder cancer diagnosis (range: 1.7–8.2%, depending on the year). Figure 1 shows the trend in secondary bladder cancer diagnosis among patients with UTUC over time. In a line-of-best-fit analysis, the correlation coefficient (r2) was equal to 0.0313 on a 0–1 scale, where r2= 1 indicates complete correlation between year of surgery and rate of secondary bladder cancer and r2= 0 indicates no correlation at all. The mean follow-up time for the patients without bladder cancer was 71 (95% CI: 62–79) months and for those with secondary malignancy it was 50 (95% CI: 48–52) months. Furthermore, there was a mean interval of 26.5 (95% CI: 22.2–30.8) months between cancer diagnoses. A total of 90% of the bladder malignancies occurred within 5 years of the initial UTUC diagnosis. Across years, similar proportions of patients died within 2.5 years of their initial UTUC diagnosis, with 49% dying before 2.5 years in the pre-1985 group and 52% dying early in the post-1985 group. Table 1 shows the baseline characteristics of the study population. There was no difference in age between the two groups with the median ages for both respective groupings being 70.9 years. Similarly, patients in both cohorts had a similar tumour grade (P= 0.138). Relative to those without secondary tumours, patients with secondary bladder malignancy were more likely to be men (P= 0.005) and non-white (P < 0.001). Furthermore, looking at upper tract tumour size as a continuous variable, those with bladder tumours were more likely to present with larger tumours (4.2 vs 3.1 cm, P < 0.001) and with tumours located in the ureter as opposed to the renal pelvis and calyces (P < 0.001). With respect to staging, the majority of patients in both cohorts had regional disease or tumour extending into adjacent peri-ureteric tissue; however, patients without a bladder recurrence more often had metastatic disease at presentation (P < 0.001), which was probably attributable to their short overall survival times after diagnosis of metastatic UTUC.

Figure 1.

Percent of all patients with UTUC who went on to have a recurrence, by year 1975–2005.

Table 1. Baseline characteristics of patients with and without a secondary bladder recurrence
CharacteristicBladder recurrenceNo bladder recurrence P
  • *

    SEER staging system: localized = subepithelial connective tissue ivaded; regional = direct extension only into adjacent tissue (i.e. peripelvic/ peri-ureteric tissue); metastatic.

N 2424970 
Tumour size, mm41.530.7<0.001
Age, years70.970.90.948
Male, n (%)166 (69)2962 (60)0.005
Married, n (%)164 (68)2995 (60%)0.019
Race/Ethnicity  0.001
 Caucasian, n (%)196 (81)4298 (86) 
 African-American, n (%)6 (2)193 (4) 
 Hispanic, n (%)9 (4)154 (3) 
 Other, n (%)31 (13)325 (7) 
Location of primary tumour  <0.001
 Kidney, n (%)19 (8)753 (15) 
 Renal pelvis, n (%)117 (44)2497 (50) 
 Ureter, n (%)106 (48)1720 (35) 
 High grade (primary), n (%)145 (60)2727 (55)0.138
SEER historic stage A*  <0.001
 Localized, n (%)94 (39)1569 (31) 
 Regional, n (%)139 (58)2216 (45%) 
 Metastatic, n (%)3 (1)853 (17) 
 Unstaged, n (%)6 (2)332 (7) 

In a multivariable model controlling for the aforementioned demographic and tumour characteristics, age, race, tumour location and tumour grade were not associated with risk of having a secondary bladder malignancy (Table 2). However, in a subset analysis examining the impact of grade, when the patients who died with 26 months of follow-up were removed from the cohort, high grade disease became a significant predictor of bladder recurrence (odds ratio [OR]: 2.02, CI: 1.18–3.45). Increased tumour size was independently associated with having a secondary cancer (OR: 0.98; 95% CI:0.97–0.99). Year of diagnosis remained an insignificant prognosticator of the secondary bladder malignancy (OR: 0.99; 95% CI: 0.95–1.03).

Table 2. Logistic regression model predicting the odds of a having a secondary bladder malignancy
CharacteristicLogistic model
OR95% CI P
Age1.010.99–1.030.30
Sex   
 MaleREF 
 Female0.830.54–1.290.42
Marital Status   
 UnmarriedREF 
 Married0.940.60–1.480.80
Race   
 CaucasianREF 
 African-American1.020.31–3.390.97
 Hispanic0.610.15–2.550.50
 Other1.590.88–2.880.13
Tumour size0.980.97–0.99<0.001
Tumour location   
 Renal pelvisREF 
 Ureter0.960.61–1.520.87
 Kidney0.800.41–1.550.50
Year of diagnosis0.980.95–1.020.41

DISCUSSION

Upper tract urothelial carcinoma continues to be a confounding disease, as evidenced by our finding that rates of secondary bladder cancer remained stagnant across the study period. The natural history of UTUC has not changed in 30 years, as secondary bladder cancer appears at a mean of just 2 years after the initital diagnosis. Large tumours, and those originating from the ureter, increase one's likelihood of having secondary bladder cancer. Surprisingly, tumour grade was unrelated to secondary cancer diagnosis; however, this finding may be attributable to the fact that half of the patients in the present analysis died within 2.5 years, suggesting that those with high grade cancer may have not lived long enough to have a secondary cancer. It is also noteworthy that >90% of patients who did have a secondary bladder malignancy were diagnosed within 5 years of their primary malignancy, a finding that may influence how urologists choose to follow up these patients over time.

It is important to interpret these findings in the context of current literature on the subject to appreciate the new findings of the present study. First, the correlations made with respect to upper tract tumour size and location are consistent with existing evidence showing that the extent of UTUC and ureteric location are significant predictors of the development of intravesical disease [6,8,9]. Indeed, although not statistically significant as an independent predictor of the development of secondary bladder cancer, a greater number of patients with bladder cancer recurrence had a ureteric UTUC location as opposed to patients without bladder malignancy (48% vs 35%, P < 0.001); this finding could be explained by both the ‘field change’ and ‘tumour seeding’ models proposed as the mechanism for UTUC spread. Also, the conclusions in the present study are supported by previous work by Ishikawa et al. [10], who recently published a retrospective review of 208 patients undergoing nephroureterectomy, 55 of whom underwent diagnostic ureteroscopy before surgery. The 2-year bladder cancer recurrence-free survival rate was similar in both groups (P= 0.972) with rates of 60.0% in the ureteroscopy group and 58.7% in controls. However, unlike the latter study, the present study involved population-level proportions and was not restricted to a cohort of patients that eventually underwent nephro-ureterectomy. Also, in contrast to the publication by Ishikawa et al., which spans 1990 to 2005, the present investigation from 1975 to 2005 is truly longitudinal, capturing an era that saw large-scale innovation in diagnostic endoscopy and treatment.

Despite the relative strengths of the present study, several limitations warrant further discussion. The overall secondary bladder recurrence rate in our study was significantly lower than the range of 15–30% previously reported in the literature [9,11,12]. There are several reasons why the rates of secondary bladder cancer may have been lower in our study population. First, the patient would only have been captured in SEER if their recurrence occurred in one of the nine SEER catchment areas. If a patient moved outside of one of these areas, the recurrence would not have been captured. Second, patients who died were not removed from the study, and have been included in the group who did not have a secondary recurrence. Last, a significant proportion of patients died within 2.5 years of their initial UTUC diagnosis (49% pre-1985 vs 52% post-1985), and although the death rate was similar in both cohorts, this phenomenon certainly diminished the perceived recurrence rates. Nevertheless, there is no evidence to support the notion that the relative rate would be significantly different across the study period owing to these factors. In fact, even if the group of patients treated after the year 2000 were removed because of inadequate follow-up, the results would not differ, since 90% of the patients in the present analysis had recurrence within 5 years of UTUC diagnosis. That being said, the advantages of looking at incidence rates in a large US representative population are somewhat mitigated by low overall estimates of secondary recurrences.

A similar observation that warrants further discussion involves the fact that tumour grade became a significant predictor of bladder recurrence when the patients who died within 26 months of follow-up were removed from the analysis. In fact, tumour grade remains a pathological predictor of the development of UTUC after primary bladder tumour diagnosis [13]. The role of grade as an independent risk factor for intravesical recurrence after UTUC identification/ management, however, has not been consistently substantiated and/ or addressed in the contemporary evaluation of this topic [8,9,14,15]. _ENREF_19 Accordingly, even patients with low grade disease have been monitored carefully for bladder recurrence. Although methods of tumour grade classification may have changed over the time span of the present study, it is important to note this feature became an independent predictor of intravesical recurrence, in accordance with some previous literature examining this phenomenon [6].

Additionally, the conclusions described in the present study must be evaluated with the following contextual and methodological considerations. First, individual information on tumour multifocality or concomitant carcinoma in situ, factors that might have a bearing on bladder recurrence rates, were not available in SEER. Second, differences in the evaluation of UTUC over the time period of this study must also be acknowledged, especially as it pertains to the increasing use of different imaging methods, CT in particular, in the current assessment of these patients [16–18]. CT improves upon the detection capability of i.v. pyelography, and this treatment method is currently a valuable tool in the staging of these tumours. The use of CT, in particular, has contributed to the earlier detection relative to the pre-imaging era when patients presented with symptoms of progressive disease [19,20]. Another chronological difference that has a bearing on these results involves the inception of laparoscopic nephro-ureterectomy in approximately 1995. Unfortunately, information on open vs laparoscopic nephro-ureterectomy is not available in SEER and cannot be assessed within the confines of the present analysis. As compared with the open technique, the degree of ureteric manipulation and distal control afforded by this approach may have affected the incidence of secondary bladder malignancy in the latter years of this study. Indeed, it has been postulated that the high pressure applied during laparoscopic surgery might facilitate tumour spread into the urinary tract, thereby promoting intravesical recurrence [13]. The conclusions with respect to this issue remain controversial, with some studies substantiating no difference in cancer-specific outcomes between the open and laparoscopic approach, while other investigations have cited laparoscopy as a risk factor for intravesical recurrence [8,14,21–24]. Because the effect of surgical approach remains equivocal it is therefore difficult to interpret the impact of this variable over the time frame of the present study.

As a final commentary, it was noteworthy that a similar number of patients throughout the study period died from UTUC within 2.5 years of follow-up, suggesting that overall survival for this cancer has not changed with time, despite corresponding advances in endoscopic techniques. This finding thereby underscores the continued importance of future work to improve both detection and management strategies for this disease.

ACKNOWLEDGEMENTS

This research was made possible by a grant from the Doris Duke Charitable foundation.

CONFLICT OF INTEREST

None declared.

Ancillary