Incidence and survival of patients with carcinoma of the ureter and renal pelvis in the USA, 1973–2005

Authors

  • Jay D. Raman,

    1. Division of Urology, Division of Outcomes Research and Quality, Department of Surgery, Penn State Milton S. Hershey Medical Center, *Penn State University College of Medicine, Division of Health Services Research, Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
    Search for more papers by this author
  • Jamie Messer,

    1. Division of Urology, Division of Outcomes Research and Quality, Department of Surgery, Penn State Milton S. Hershey Medical Center, *Penn State University College of Medicine, Division of Health Services Research, Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
    Search for more papers by this author
  • John A. Sielatycki,

    1. Division of Urology, Division of Outcomes Research and Quality, Department of Surgery, Penn State Milton S. Hershey Medical Center, *Penn State University College of Medicine, Division of Health Services Research, Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
    Search for more papers by this author
  • Christopher S. Hollenbeak

    1. Division of Urology, Division of Outcomes Research and Quality, Department of Surgery, Penn State Milton S. Hershey Medical Center, *Penn State University College of Medicine, Division of Health Services Research, Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
    Search for more papers by this author

Jay D. Raman, Division of Urology, Penn State Milton S. Hershey Medical Center, 500 University Drive, MC H055, Hershey, PA 17033, USA.
e-mail: jraman@hmc.psu.edu

Abstract

Study Type – Prognosis (retrospective cohort)
Level of Evidence 2b

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

Upper-tract urothelial carcinoma (UTUC) is a relatively uncommon urological malignancy with survival and outcomes data largely determined from single-centre series which can be limited by relatively small case numbers.

Through review of a large population based cohort, this study provides valuable information regarding epidemiological and survival patterns for over 13,000 patients with UTUC diagnosed over the past three decades.

OBJECTIVE

• To evaluate epidemiological and survival patterns of upper-tract urothelial carcinoma (UTUC) over the past 30 years through a review of a large, population-based database.

PATIENTS AND METHODS

• Data from the Surveillance, Epidemiology and End Results (SEER) database from 1973 to 2005 were reviewed in 10-year increments to evaluate disease trends.

• Univariate and multivariate survival analyses identified prognostic variables for outcomes.

RESULTS

• In total, 13 800 SEER-registered cases of UTUC were included. The overall incidence of UTUC increased from 1.88 to 2.06 cases per 100 000 person-years during the period studied, with an associated increase in ureteral disease (0.69 to 0.91) and a decrease in renal pelvic cancers (1.19 to 1.15).

• The proportion of in situ tumours increased from 7.2% to 31.0% (P < 0.001), whereas local tumours declined from 50.4% to 23.6% (P < 0.001).

• There was no change in the proportion of patients presenting with distant disease.

• In multivariate analysis, increasing patient age (P < 0.001), male gender (P < 0.001), black non-Hispanic race (P < 0.001), bilateral UTUC (P= 0.001) and regional/distant disease (P < 0.001) were all associated with poorer survival outcomes.

CONCLUSIONS

• The incidence of UTUC has slowly risen over the past 30 years.

• Increased use of bladder cancer surveillance regimens and improved abdominal cross-sectional imaging may contribute to the observed stage migration towards more in situ lesions.

• Although pathological disease characteristics impact cancer outcomes, certain sociodemographic factors also appear to portend worse prognosis.

Abbreviations
SEER

Surveillance, Epidemiology and End Results

URS

ureteroscopy

UTUC

upper-tract urothelial carcinoma.

INTRODUCTION

Upper-tract urothelial carcinoma (UTUC), including tumours of the ureter and renal pelvis, is an uncommon genitourinary malignancy accounting for 5% of urothelial cancers and less than 10% of renal tumours [1]. Owing to its infrequent presentation, trends in disease incidence, associated demographic factors and predictors for cancer-related and overall survival outcomes are limited. Most contemporary data originate from single-institution series, which are hindered by small sample sizes and homogeneity in study populations. Data compiled from multiple institutions have successfully addressed some of these shortcomings, although such studies may also be limited by the fact that they originate from tertiary centres of excellence and reflect a primary surgical population [2].

The Surveillance, Epidemiology, and End Results (SEER) programme of the National Cancer Institute accrues data on cancer incidence and survival for 26% of the US population. Munoz and Ellison [3] previously used this large, population based database to identify incidence trends for UTUC between 1973 and 1996. Subsequently, however, the landscape of urological cancer care has evolved. Namely, there have been improvements in axial abdominal imaging (CT and/or MRI), the discovery of novel urinary cancer markers, a refinement of endoscopic/ureteroscopic instrumentation and the introduction of chemotherapeutic regimens, resulting in prolonged survival for bladder cancer patients [4–7]. Collectively, these variables are likely to continue to impact disease trends and outcomes for upper-tract urothelial neoplasms.

Therefore, with SEER data now available up to 2005, the present study sought to evaluate the epidemiological and survival patterns for UTUC diagnosed over the past 30 years in the USA.

MATERIALS AND METHODS

DATA

The SEER public database (http://www.seer.cancer.gov) was used as the data source for the present study. Incident cancer cases have associated demographic and pathological data available, whereas subsequent treatment and recurrence data are not recorded.

Comparing data from 1973 to 2005 required the use of the SEER 9 registry (Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco–Oakland, Seattle–Puget Sound, and Utah), which comprises 10% of the US population. Cases were associated with population data using three racial groups: white, black and ‘other’. The ‘other’ race category used in the SEER 9 Registry database consists of American Indian/Alaska Native and Asian/Pacific Islander combined.

The SEER 9 registry from 1973 to 2005 was queried for renal pelvic and ureteral tumours as designated by the International Classification of Diseases codes ICD-O-3 C65.9 (renal pelvis) and C66.9 (ureter) [8]. The analysis identified 13 800 patients of whom 12 491 were white (including Hispanic), 519 were black and 790 were of ‘other’ race/ethnicity. Some 8600 of those diagnosed were male; 5200 were female. Comparisons across the time periods 1973–1984, 1985–1996 and 1997–2005 were performed with respect to demographic variables (age, gender, race/ethnicity), disease stage at diagnosis (in situ, local, distant, regional) and overall survival. The absence of linked Medicare data precludes the direct assessment of cancer-specific survival; thus, the endpoint within the present study was overall survival (all-cause mortality). Race/ethnicity was categorized into four groups: White non-Hispanic (WNH), Black non-Hispanic (BNH), Hispanic and other. Using the full cohort of patients, which was age-standardized to the US 2000 census, incidence rates of UTUC (cases per 100 000 person/year) were computed.

To permit comparisons across three decades of data coded in the SEER database, it was necessary to utilize the SEER 2000 summary staging system [9]. Here, in situ tumours are defined as intraepithelial, non-infiltrating cancers that do not invade the basement membrane of the organ. A localized tumour is confined to the organ of origin without extension beyond the primary organ. Regional extension of tumour included direct extension to adjacent organs or structures or spread to regional lymph nodes. Finally, if the cancer had spread to parts of the body remote from the primary tumour, these were classified as distant stage.

STATISTICAL ANALYSIS

Statistical analysis was performed using STATA software, version 9.0 (Stata Corp., College Station, TX, USA). Comparisons were tested using anova for continuous variables and chi-squared tests for binary and categorical variables. Survivor functions were estimated, stratified by several variables of interest, using the Kaplan–Meier product limit method. Log-rank tests were used to compare survivor functions. Cox proportional hazards modelling estimated the risk of survival controlling for various patient and disease factors, including age, gender, race/ethnicity and stage of disease. Univariate survival analysis suggested that the proportional hazards assumption for disease site (ureter vs renal pelvis) was not met; therefore, separate multivariate models were constructed for each tumour location. P≤ 0.05 (two-sided) was considered statistically significant.

RESULTS

Between 1973 and 2005, there were 13 800 SEER-registered cases of upper-tract urothelial neoplasms. Figure 1 presents the age-adjusted incidence rates of renal pelvic disease, disease of the ureter and total combined incidence. Over the past 30 years, the incidence of disease in the ureter increased from 0.69 to 0.91 cases per 100 000 person-years, renal pelvic disease incidence decreased slightly from 1.19 to 1.15 cases per 100 000 person-years, and the total incidence of upper-tract tumours rose from 1.88 to 2.06 cases per 100 000 person-years.

Figure 1.

Incidence of upper-tract urothelial carcinoma (ureter and renal pelvis) age-adjusted to the US 2000 census population.

Table 1 highlights demographic and pathological characteristics for treated patients stratified over three time periods: 1973–1984, 1985–1996 and 1997–2005. The mean age at diagnosis increased each 10-year increment, with an overall increase of 5 years (68 to 73 years) over the entire study period. In addition, there were changes in race/ethnicity and gender distribution, with more non-whites and females diagnosed at later time points. Perhaps, most importantly, the proportion of in situ tumours continued to increase at each time increment (7.2% to 23.0% to 31.0%), with an associated 50% reduction in the incidence of local tumours over the study period (50.4% to 23.6%, P < 0.001). UTUC regional disease increased by 2.6% from the first to third decade studied (P= 0.003), whereas the incidence of distant disease did not change over time (P= 0.12).

Table 1.  Characteristics of patients with upper-tract urothelial carcinoma (Surveillance, Epidemiology and End Results, 1973–2005)
Variable1973–1984(n= 4064)1985–1996(n= 5122)1997–2005(n= 4614)P
Age (years)68.070.073.00.004
Gender (%)    
 Male64.161.961.30.02
 Female35.938.238.70.02
Race (%)    
 White92.690.988.3<0.001
 Black3.43.64.30.09
 Other4.05.57.5<0.001
Stage (%)    
 In situ7.223.031.0<0.001
 Local50.431.923.6<0.001
 Regional33.737.136.30.003
 Distant8.78.09.20.12

Survival curves stratified by clinical characteristics are presented in Figs 2–4. As shown in Fig. 2, increasing age at the time of diagnosis was significantly associated with poorer survival outcomes (P < 0.001). For example, 5-year overall survival decreased from 75% for patients <50 years of age to 50% for patients aged 70–79 years and <20% for those aged ≥85 years. In addition, male gender was associated with worse overall survival (P < 0.005), with the magnitude of absolute difference being 7% (64% vs 57%) and 5% (52 vs 47%) at 3 and 5 years, respectively (Fig. 3). Differences were also observed in the survival of UTUC patients when stratified by race/ethnic background (Fig. 4). Specifically, BNH patients appeared to have poorer survival outcomes at both 5 and 10 years compared to WNH or ‘other’ patients. Interestingly, compared to the BNH cohort, Hispanic patients had an improved 5-year survival, although 10-year survival was 30% in both groups, suggesting competing causes of mortality may contribute to the lower survival in these two cohorts.

Figure 2.

Survival analysis of patients with upper-tract urothelial carcinoma stratified by age. Dx, diagnosis.

Figure 3.

Survival analysis of patients with upper-tract urothelial carcinoma stratified by gender. Dx, diagnosis.

Figure 4.

Survival analysis of patients with upper-tract urothelial carcinoma stratified by race/ethnicity. BNH, black non-Hispanic; WNH, white Non-Hispanic; Dx, diagnosis.

Figures 5 and 6 present the survival outcomes as stratified by pathological characteristics. As expected, patients with in situ UTUC lesions showed the best overall survival with rates worsening with advancing tumour stage (Fig. 5). Patients with distant metastases appeared to have an ominous prognosis, with survival rates under 10% some 3 years after diagnosis. Irrespective of laterality, survival for patients with known unilateral disease was similar. However, patients with bilateral disease had significantly worse outcomes (P < 0.001) (Fig. 6). Interestingly, patients with unspecified laterality had poorer outcomes than either left- or right-sided disease, suggesting this cohort may potentially include unrecorded bilateral cases.

Figure 5.

Survival analysis of patients with upper- tract urothelial carcinoma stratified by stage (Surveillance, Epidemiology and End Results, 2000; summary staging system). Dx, diagnosis.

Figure 6.

Survival analysis of patients with upper-tract urothelial carcinoma stratified by laterality. Dx, diagnosis.

Separate Cox proportional hazards multivariate models were fit for patients with disease in the ureter (Table 2) and renal pelvis (Table 3). For ureteral urothelial cancers, the risk of mortality was significantly associated with increasing age of the patient. Furthermore, men with ureteral cancer had a 16% greater risk of death compared to women. When considering race/ethnicity, BNH patients had an increased risk of mortality (hazard ratio, HR, 1.33, P= 0.006) compared to WNH (reference population). However, no differences in mortality were observed for Hispanic patients (HR, 1.01; P= 0.48) or patients of other race/ethnicity (HR, 0.89; P= 0.12) compared to WNH patients. Compared to patients with localized tumours, those with in situ lesions had 23% lower risk of death (P < 0.001), whereas regional and distant disease increased the mortality risk by 1.7- and 6.1-fold, respectively. Finally, patients with bilateral UTUC had a significantly greater risk of death compared to those with unilateral upper-tract disease (HR, 1.79; P= 0.001).

Table 2. 
Results for multivariate survival analysis of patients with disease of the ureter
VariableHazardratio95% CIP
LowerUpper
Age (years)    
 18–40Reference   
 41–501.910.983.700.06
 51–602.841.515.350.001
 61–704.852.599.08<0.001
 71–808.064.3115.09<0.001
 ≥8112.306.5523.09<0.001
Gender    
 FemaleReference   
 Male1.161.081.24<0.001
Race/ethnicity    
 Other0.890.781.030.12
 Black non-Hispanic1.331.091.630.006
 Hispanic1.010.971.060.48
 White non-HispanicReference   
Stage    
 In situ0.880.810.960.006
 LocalReference   
 Regional1.701.581.84<0.001
 Distant6.145.387.02<0.001
Laterality    
 LeftReference   
 Right0.970.911.040.40
 Bilateral1.791.272.510.001
Table 3. 
Results for multivariate survival analysis of patients with disease of the renal pelvis
VariableHazardratio95% CIP
LowerUpper
Age (years)    
 18–40Reference   
 41–501.511.102.090.01
 51–602.581.923.46<0.001
 61–704.093.065.47<0.001
 71–806.354.758.49<0.001
 ≥819.647.1812.94<0.001
Gender    
 FemaleReference   
 Male1.111.051.18<0.001
Race/ethnicity    
 Other0.800.700.90<0.001
 Black non-Hispanic1.291.121.47<0.001
 Hispanic0.990.961.030.74
 White non-HispanicReference   
Stage    
 In situ0.770.700.85<0.001
 LocalReference   
 Regional1.631.531.74<0.001
 Distant6.776.177.42<0.001
Laterality    
 LeftReference   
 Right0.990.741.320.94
 Bilateral1.030.981.090.26

Among patients with urothelial cancer of the renal pelvis, similar observations were noted (Table 3). Increasing patient age (P < 0.001), male gender (P < 0.001), black non-Hispanic race (P < 0.001) and regional and distant disease (P < 0.001 for both) were all associated with poorer survival outcomes. Interestingly, as opposed to ureteral cancer, bilateral renal pelvis disease was not associated with poorer survival (HR, 1.03, P= 0.26).

DISCUSSION

As a result of its infrequent presentation, trends in outcomes and survival for UTUC are incompletely understood, with most data originating from older, population-based studies. Mellemgaard et al.[10] reported on disease trends observed from the Danish cancer registry over a 40-year period (1943–1988) and noted a more than tenfold increase in the incidence of UTUC neoplasms over that time [10]. In 2000, Munoz and Ellison [3] reviewed over 20 years of data from the USA and similarly noted an increase in UTUC incidence, although this was of lower magnitude than the Danish study. In this latter study, trends were observed towards diagnosis of earlier stage lesions. Since the mid-1990s, however, changes in diagnostic and treatment modalities related to UTUC have occurred. It was hypothesized that disease trends would continue to evolve when considering a more contemporary patient population.

An increase in the overall incidence of UTUC was observed over the past three decades (Fig. 1). This trend was driven by a steady rise in the incidence of ureteral neoplasms (0.69 to 0.91 cases per 100 000 years), whereas renal pelvic disease remained constant over the study period. Accompanying this increase in UTUC malignancies was a trend towards diagnosis of earlier stage lesions. Specifically, an increased detection of in situ lesions from 7.2% in the first decade to 31.0% in the third decade was noted in the present study. This increase was paralleled with a corresponding decline in local UTUC tumours, such that the combined incidence of in situ and local tumours remained fairly stable (57.6%, 54.9%, and 54.6% in the first, second and third decades, respectively). No appreciable differences were observed in the incidence of distant disease over time, with a slight, albeit significant, increase in regional UTUC (Table 1). Finally, with respect to demographic data, the age of patients diagnosed with UTUC increased with each 10-year increment, with an overall increase of 5 years (68 to 73 years) over the study period.

It is hypothesized that these observations probably centres around three variables: improved bladder cancer survival, refinements in cross-sectional imaging, and newer endoscopic technology. Historically, patients with bladder cancer have experienced poor overall survival. However, refinements in surgical technique and targeted chemotherapy agents have translated to improved bladder cancer outcomes [7,11,12]. This probably is most pronounced in patients with superficial bladder cancer and a review by Herr et al.[13] implicates only modest gains in survival for pT2–pT4 malignancies treated by radical cystectomy stratified before and after 1985. Accompanying this patient longevity is an increased susceptibility to developing metachronous upper-tract lesions [14,15]. Tran et al.[14] highlighted this risk by noting that the 3- and 5-year cumulative incidence of UTUC following bladder cancer was 4% and 7%, respectively. Thus, more early stage tumours in older patients may be diagnosed as a result of upper-tract surveillance in surviving bladder cancer patients.

Improvements in imaging also probably contribute to the increased detection of in situ lesions. Historically, upper-tract imaging utilized IVU. However, contemporary axial imaging with CT or MR urography improves the detection of small lesions compared to IVU [4]. A recent meta-analysis of published literature on CT urography highlights a pooled sensitivity 96% and a pooled specificity of 99%[16]. This compares very favourably with the reported sensitivities for excretory urography, which are in the range 50–60%[17,18]. An additional technological advance in the past three decades has been improvements in ureteroscopy (URS). Although URS was initially introduced in the 1950s, the development of flexible endoscopes in the mid-1980s improved access to the upper ureter and renal pelvis. Subsequent refinements with digital ureteroscopes and the incorporation of narrow band imaging have further allowed the identification of early stage lesions [19–21]. Studies suggest that the addition of URS improves the diagnostic accuracy for UTUC by 85–90% compared with excretory urography alone [22]. In addition, URS allows for endoscopic evaluation and biopsy of suspicious areas with good histological correlation with final pathological grade [23]. Collectively, these factors all have probably promoted a stage migration to earlier tumours.

In addition to evaluating disease trends, predictors for disease outcomes were also identified. Both pathological (stage, laterality) and demographic (age, gender, race/ethnicity) variables were associated with UTUC outcomes (Tables 2,3). As expected, stage at diagnosis strongly correlated with overall survival in the cohort analyzed. These findings are similar to observations made from a recently published large multi-institutional cohort of patients with UTUC treated by radical nephroureterectomy [2]. Although the SEER database precludes a more detailed analysis of additional pathological variables, contemporary data indicate that tumour grade, lymph node metastases, sessile architecture and lymphovascular invasion are also associated with UTUC outcomes [2].

Age correlated strongly with overall survival. Five-year overall survival decreased from 75% for patients <50 years of age to <20% for those aged ≥85 years. These observations are probably attributable to older patients being less likely to undergo definitive surgical resection, less able to tolerate chemotherapy regimens, and dying of competing causes of mortality [24]. Inman et al.[25] highlighted this latter point by noting that competing risks were responsible for 46% of deaths in a cohort of 168 patients with surgically treated UTUC. Race/ethnicity was associated with differences in overall survival, with BNH patients experiencing a 30% higher mortality rate than Hispanic or WNH (Tables 2,3). It is hypothesized that these observed differences may be related to genetic and biological differences, access to health care, and baseline health status. Similar racial/ethnic disparities have been noted for other urological malignancies and, unfortunately, they appear to persist even when patients are enrolled in clinical trials with standardized regimens [26–28]. Finally, although differences in survival based on gender were noted, it is difficult to assess whether this is a result of the biology of disease or the presence of associated comorbid conditions in men. Underwood et al.[29] found a similar gender disparity in bladder cancer outcomes when reviewing 25 years of SEER data. More recently, however, Fernandez et al.[30] failed to appreciate any differences in clinicopathological features or prognosis between men and women treated with radical nephroureterectomy for UTUC.

There are limitations to the present study. First, comparison across decades necessitated utilizing the SEER summary staging system. In this regard, pathologically distinct entities as appreciated by the TNM [31] system may be grouped together in the SEER system. For example, in situ tumours include both Ta and Tis lesions, which probably have variable behaviours. Second, in the absence of linked Medicare claims data, mortality from competing risks is difficult to quantify. Thus, the specific determination of recurrence-free or cancer-specific survival outcome measures was not possible. Currently, we are conducting a review of linked SEER Medicare data with the goal of better delineating death from disease vs that attributable to competing causes of mortality. Third, the data of the present study fail to consider the significance of synchronous or metachronous bladder cancer. Indeed, an upper-tract recurrence following bladder cancer may behave differently than a de novo UTUC lesion. Despite these limitations, for this relatively uncommon neoplasm, it is considered that the SEER database presents an extensive cohort sample for evaluating national disease trends.

In conclusion, the incidence of UTUC has increased over the past 30 years, with an associated stage migration towards earlier stage tumours. Nonetheless, over 40% of patients continue to present with regional or distant disease. Such observations emphasize the need for improved tools to identify or screen patients at risk for developing UTUC, as well as the need for improved therapies to manage advanced UTUC. Although pathological characteristics impact UTUC outcomes, sociodemographic factors, including age, race/ethnicity and gender, are also associated with a worse prognosis.

CONFLICT OF INTEREST

None declared.

Ancillary