Presented in part at the annual meeting of the American Urological Association, Atlanta, Georgia, May 20-25, 2006.
Surveillance of urothelial carcinoma†
Stage and grade migration, 1993–2005 and survival trends, 1993–2000
Article first published online: 12 FEB 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 7, pages 1435–1447, 1 April 2009
How to Cite
David, K. A., Mallin, K., Milowsky, M. I., Ritchey, J., Carroll, P. R. and Nanus, D. M. (2009), Surveillance of urothelial carcinoma. Cancer, 115: 1435–1447. doi: 10.1002/cncr.24147
- Issue published online: 19 MAR 2009
- Article first published online: 12 FEB 2009
- Manuscript Accepted: 15 SEP 2008
- Manuscript Revised: 25 AUG 2008
- Manuscript Received: 4 JUN 2008
- Empire Clinical Research Investigator Program
- urothelial carcinoma;
- stage migration;
Previous investigators have detected shifts to lower stages at diagnosis for renal cell carcinoma and prostate cancer. The authors investigated whether a similar pattern is seen for urothelial carcinomas of the bladder, ureter, and renal pelvis and sought to identify changes in cancer grade and survival trends from 1993 to 2005.
The National Cancer Data Base (NCDB) collects data on approximately 75% of all newly diagnosed cancer cases annually. The authors queried the database for cases of urothelial carcinomas diagnosed in 1993-2005 in patients aged 18 years and older. All cancer stage data were forward converted to the sixth edition of the American Joint Committee on Cancer staging definitions.
A total of 334,480 bladder cancer cases, 15,105 renal pelvis cancer cases, and 10,128 ureteral carcinoma cases were identified. Stage data were available for 84% of bladder cases, 83% of renal pelvis cases, and 82% of ureter cases. With the classification of early stage tumors as stage 0a, 0is, and I and late stage tumors as II, III, and IV, the percentage of early stage renal pelvis and ureter tumors increased slightly from 1993 to 2005, whereas no stage migration was seen in bladder tumors. In looking specifically at early stage tumors, a significant increase in the proportion of stage 0a and a significant decrease in the proportion of stage I tumors for each cancer site was seen between 1993 and 2005. The proportion of high grade tumors in each disease site significantly increased from 1993 to 2005. For cases diagnosed in 1993-1996 and 1997-2000, a significant decrease in 5-year relative survival was observed for patients with stage I and stage II bladder cancer. The absolute change, however, was relatively small, and for bladder cases was not significantly different when adjusted for low or high grade tumors.
When differentiating between early and late stage tumors, a slight stage migration was seen in renal pelvis and ureteral carcinomas, whereas no stage migration was seen in bladder tumors. Within early stage tumors of all sites, a stage shift was seen, most notably with the proportion of stage 0a tumors increasing and stage I tumors decreasing. The proportion of high grade tumors in all sites increased. No change in overall survival was observed, underscoring the need for new therapeutic advances. Cancer 2009. © 2009 American Cancer Society.
Urothelial carcinomas include bladder and upper urinary tract cancers involving the renal pelvis and ureters. Estimated new cases of bladder cancer for 2007 in the United States totaled 67,160; among American men, it is the fourth most common new cancer.1 Upper urinary tract neoplasms are less common and represent only 5% of all urothelial tumors.2 Munoz and Ellison estimated the incidence of ureteral transitional cell carcinomas (TCC) at 0.73 of 100,000 person-years and that of renal pelvis TCC at 1.0 of 100,000 person-years between 1985 and 1995.3
Recent studies of certain genitourinary malignancies have suggested a stage migration, ie, a trend toward a lower disease stage at the time of diagnosis. This phenomenon has been most pronounced in prostate cancer and has been noted in renal cell carcinoma (RCC), as well.4-11 Factors underlying the development of stage migrations include screening tests, such as the prostate-specific antigen in prostate cancer and the use of imaging that can uncover asymptomatic early stage tumors, such as in RCC. An analysis by Kane et al of patients with RCC presenting between 1993 and 2004 linked a small survival improvement to an increase in earlier stage disease, possibly owing to an increase in incidentally found tumors by imaging.12
We sought to determine whether a similar stage migration occurred in bladder and upper urinary tract urothelial carcinomas between 1993 and 2005 and analyzed trends in grade at diagnosis as well. In addition, we also analyzed survival trends to determine whether hypothesized stage and grade migrations could be associated with changes in survival, as have been seen in other malignancies.
MATERIALS AND METHODS
The National Cancer Data Base (NCDB), a hospital-based longitudinal information source, was established in 1989 as a joint project of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society to serve as a comprehensive clinical surveillance resource for cancer care in the United States. It has been described in detail elsewhere.13 The NCDB currently captures approximately 75% of all newly diagnosed cancer cases from over 1400 facility-based cancer registries annually and holds information on over 19 million cases of cancer diagnosed between 1985 and 2005.
The NCDB was queried for cases of bladder, ureteral, and renal pelvis urothelial carcinoma diagnosed between 1993 and 2005. Cases were limited to the following World Health Organization (WHO) International Classification of Disease for Oncology, third edition (ICD-O-3) histology codes that define transitional cell carcinoma in patients aged 18 years and older: 8120 (transitional cell carcinoma), 8122 (spindle cell transitional cell carcinoma), and 8130 (papillary transitional cell carcinoma).14 From 1993 to 2003, American Joint Committee on Cancer (AJCC) staging definitions for bladder cancer were consistent in the third and fourth editions (1993-1997), but had definitional changes in the fifth and sixth editions (1998-2005). To compare the stage data over the 13-year period, all of the data were forward converted to AJCC sixth edition. Cases before 1993 were not used because of definitional changes in bladder cancer AJCC T and N categories between the second and third edition.
From the third to sixth AJCC editions, no major changes occurred in the staging definitions for stages T0, Ta, Tis, T1, T4a (stage III), and the collapsed T4 group (stage IV T4, T4b). The third and fourth edition definition of T2 is the following: the tumor invades the superficial muscle (inner half). In the fifth and sixth editions, this definition describes the T2a group. The T2 sixth edition definition is now more inclusive: the tumor invades muscle. Those cancers that were staged as T2 using the third and fourth editions were moved to the T2 group for this analysis. Those recorded as T3a in the third and fourth editions had the following definition: tumor invades deep muscle (outer half). These cases were also converted to the fifth and sixth editions T2 group. The T3 group includes those cases staged as T3 in the fifth and sixth editions and those recorded as T3b in the third and fourth editions based on the following definition: tumor invades perivesical fat. The cases recorded as T3 (without subclassification as T3a or T3b) in the third and fourth editions have been removed from analysis because the third and fourth editions define these as the following: tumor invades deep muscle or perivesical fat. No way exists to break apart this group to fall into the separated categories of the fifth and sixth editions (T2: tumor invades muscle or T3: tumor invades perivesical tissue). No changes to the regional lymph nodes (N) and distant metastases (M) categories were made between the editions of interest.
When the TNM edition number was given as not staged, first edition, second edition, not applicable, or unknown, these cases were staged according to the AJCC rules for the corresponding diagnosis year. AJCC sixth edition best stage was created using a combination of the cleaned clinically and pathologically staged cases for bladder cancers. To create best T and N staging components, pathological data were taken over clinical data, and if pathological data were missing or unknown, clinical data were used. For the best M variable, clinical data were taken over pathological data, and if the clinical data were missing or unknown, pathological data were used.
Renal pelvis and ureteral cases did not have major changes to the staging definitions from the third to sixth AJCC editions. The T, N, and M variables for both clinical and pathological stages were used to create best stage for both cancer sites using the same strategy as outlined above for bladder cancers. Table 1 depicts the correlation between the T, N, M staging criteria, the AJCC classifications, muscle invasiveness (for bladder cancer), and the early/late stage classification.
|Location||T, N, M Criteria||AJCC Stage||Early/Late||Muscle Invasiveness|
|Bladder||Ta N0 M0||Stage 0a||Early||Not muscle invasive|
|Tis N0 M0||Stage 0is|
|T1 N0 M0||Stage I|
|T2 N0 M0||Stage II||Late||Muscle invasive|
|T3a N0 M0||Stage III|
|T3b N0 M0|
|T4a N0 M0|
|T4b N0 M0||Stage IV|
|Any T, N1 M0|
|Any T, N2 M0|
|Any T, N3 M0|
|Any T, any N M1|
|Renal pelvis/ureter||Ta N0 M0||Stage 0a||Early||Not applicable|
|Tis N0 M0||Stage 0is|
|T1 N0 M0||Stage I|
|T2 N0 M0||Stage II||Late|
|T3 N0 M0||Stage III|
|T4 N0 M0||Stage IV|
|Any T N1 M0|
|Any T N2 M0|
|Any T N3 M0|
|Any T, any N M1|
Trends were assessed for each individual stage and for early stage (stages 0a, 0is, and I), and late stage (stages II, III, and IV). Histologic grade was grouped into low grade (well differentiated, moderately differentiated) and high grade (poorly differentiated, undifferentiated) for comparison of percentage distribution over the 13-year period and for survival analysis. Grading was recorded as defined by the ICD-O-314 and Facility Oncology Registry Data Standards (FORDS) manual.15 Of note, the WHO grading criteria, which are different, were not used. Cases in which ICD-O/FORDS grading data were not available or in which only WHO grading criteria were reported were classified as having missing data. Overall, approximately 11% of cases had grade data classified as such.
Significance of trends in stage and grade distribution between 1993 and 2005 was assessed with the Cochran-Armitage chi-square test for trend. Differences in the percentage and mean distribution of demographic variables were compared for grouped diagnosis years 1993-1996, 1997-2000, and 2001-2005 using chi-square and Student t test. A significance level of .05 was used. Multivariate Cox regression was used to assess the risk of mortality between 1993-1996 and 1997-2000 adjusting for sex, age, race, grade, and stage, using a stepwise forward model with a .05 value for entry and .10 value for removal. SPSS version 14.0 was used for these analyses.
Because cause of death is not available, relative 5-year survival rates and corresponding 95% confidence intervals were calculated by stage and grade for cases diagnosed in 1993-1996 and 1997-2000. Relative survival is the ratio of the observed survival rate to the expected survival rate adjusted for age, sex, race, and Hispanic origin. Expected survival rates are based on 1990 life expectancy tables from the National Cancer Institute.16 Two-sided 95% confidence intervals were calculated.
Data reported to the NCDB are retrospective in nature, without patient or physician identifiers. The American College of Surgeons has executed a Business Associate Agreement that includes a data-use agreement with each of its CoC-approved hospitals. Results reported in this study were in compliance with the privacy requirements of the Health Insurance Portability and Accountability Act of 1996 as reported in the Standards for Privacy of Individually Identifiable Health Information; Final Rule (45 CFR Parts 160 and 164).
Of the 398,125 newly diagnosed bladder cancer cases in 1993-2005, 16% (n = 63,645) could not be staged because of missing or invalid M (n = 46,020), N (n = 56,442), or T (n = 20,218). Of the 18,172 newly diagnosed renal pelvis cases, 3067 (17%) could not be staged because of missing or invalid M (n = 2030), N (n = 2602), or T (n = 1507). Among the 10,128 ureteral cases, 2272 (18%) could not be staged because of missing or invalid M (n = 1527), N (n = 1972), or T (n = 1087) staging information.
A total of 334,480 bladder cancer cases, 15,105 renal pelvis cancer cases, and 10,128 ureteral carcinoma cases could be staged. Table 2 illustrates baseline demographic data for these patients. In all 3 time periods—1993-1996, 1997-2000, and 2001-2005—the majority of patients were older than 60 years, white, and male. The mean age of 70.0 years in 2001-2005 was essentially the same as that in earlier diagnosis years (1997-2000: 69.2 years; 1993-1996: 69.5 years), and there was a significant increase in the percentage of patients aged 80 years and older between 1993-1996 (19.4%) and 2001-2005 (23.2%) (trend test, P < .0001).
|Variables||Diagnosis Years||No. of Patients|
|Age group, %|
|Mean age, y||69.2||69.5||70.0|
|No. of patients||99,621||120,465||139,627||359,713|
In comparing the proportions of early stage disease in 1993 and 2005, a slight increase in early stage renal pelvis cancer was seen: 44% in 1993 to 48% in 2005 (trend test, P = .006) (Fig. 1). The proportion of early stage ureteral carcinomas increased only slightly from 52% in 1993 to 57% in 2005 (trend test, P = .08) (Fig. 2). No significant change was seen in the proportion of early stage bladder cancers, 74% in both 1993 and 2005 (Fig. 3). These trends were similar among all 3 hospital subtypes present in the database–community cancer center, comprehensive community cancer center, and academic/tertiary care cancer center (data not shown). Thus, a potential bias introduced by differences in the types of treatment facilities does not appear to be present.
Although the proportion of combined early stage bladder cancers (stage 0-I) did not change during the study period, a significant increase in the proportion of stage 0a disease from 30.1% in 1993 to 46.4% in 2005 was seen (trend test, P < .001). Countering this trend were significant decreases in the proportions of stage 0is and stage I bladder cancers (Fig. 3 and Table 3). These trends were analyzed in patients with primary tumors solely.
The increase in early stage renal pelvis cancers is mainly accounted for by the increase in the proportion of stage 0a cases, from 10.5% in 1993 to 23.2% in 2005 (trend test, P < .001) and the decrease in the proportion of stage II cases from 18.7% to 8.7% (trend test, P < .001) (Fig. 1 and Table 4). Similarly, the increase in early stage ureteral carcinomas is accounted for mainly by the increase in the proportion of stage 0a cases—12.7% in 1993 compared with 30.7% in 2005 (trend test, P < .01)—and the decrease in the proportion of stage II cases from 21.9% to 16.1% (trend test, P <.001) (Fig. 2 and Table 5). Although the proportion of early stage cases as a whole in both renal pelvis and ureteral cancers increased during this time period, the proportion of stage I cases notably decreased in both disease sites, as well as in bladder cancer (Figs. 1-3).
The percentage of stage II, III, and IV bladder tumors changed by less than 1% between 1993 and 2005. The decrease in stage II renal pelvis and ureter tumors was accompanied by an increase in stage III tumors for renal pelvis (7% increase; trend test, P < .001) and for ureter (2% increase; trend test, P < .01)(Figs. 1 and 2).
The proportion of high grade tumors also showed significant increases for each cancer site (test for trend, P < .0001) (Table 6). The proportion of high grade bladder tumors increased from 38.2% in 1993 to 49.6% in 2005. The proportion of high grade renal pelvis tumors increased from 46.8% in 1993 to 57.8% in 2005, and the proportion of high grade ureteral carcinomas increased from 44.5% in 1993 to 53.0% in 2005 (Fig. 4). When cases were divided into early and late stages, the trend toward an increasing proportion of high grade tumors was still apparent in both groups (data not shown). Not surprisingly, although the proportion of high grade tumors increased in both early and late stage cancers, the overall majority of early stage cases comprised low grade tumors, whereas late stage cases overwhelmingly comprised high grade tumors.
|Cancer Site||Diagnosis Years|
Five-year relative survival rates by grouped diagnosis years (1993-1996 vs 1997-2000), stage, and cancer site are presented in Table 7. Between these 2 time periods, a small but statistically significant survival decrease occurred for patients with stage I and stage II bladder cancer. In all other stages and overall, no significant change in bladder cancer survival was seen. As expected, 5-year relative survival decreased with increasing stage at diagnosis. When patients were subclassified by stage (early vs late) and by grade (low vs high), no significant change in bladder cancer survival was seen between the 2 time periods (Table 8).
|Stage||Diagnosis Years 1993-1996||Diagnosis Years 1997-2000|
|Relative Survival||95% Confidence Intervals||No.||Relative Survival||95% Confidence Intervals||No.|
|Location||Stage/Grade||Diagnosis Years 1993-1996||Diagnosis Years 1997-2000|
|Relative Survival||95% Confidence Intervals||No.||Relative Survival||95% Confidence Intervals||No.|
|Bladder||Early stage, low grade||98.1||97.6-98.6||48,784||97.6||97.0-98.1||57,663|
|Early stage, high grade||85.6||84.4-86.8||14,110||85.9||84.9-86.9||19,988|
|Late stage, low grade||58.4||56.0-60.8||3465||54.6||52.2-57.1||3479|
|Late stage, high grade||42.5||41.5-43.5||18,562||41.3||40.4-42.2||23,142|
|Renal pelvis||Early stage, low grade||95.5||92.4-98.6||1378||89.6||86.5-92.7||1644|
|Early stage, high grade||85.4||79.1-91.8||443||87.3||82.1-92.4||667|
|Late stage, low grade||73.5||68.5-78.6||689||62.7||56.9-68.5||593|
|Late stage, high grade||38.5||35.1-41.8||1367||36.8||33.6-40.0||1582|
|Ureter||Early stage, low grade||90.7||86.8-94.6||1072||95.8||92.2-99.3||1316|
|Early stage, high grade||68.8||59.8-77.9||257||82.1||74.2-90.0||351|
|Late stage, low grade||63.7||56.0-71.4||344||65.7||58.1-73.2||352|
|Late stage, high grade||38.9||34.4-43.3||817||39.2||35.0-43.3||1012|
Among renal pelvis cases, a 5% decline in survival occurred among stage 0a and a 6% decline occurred in stage I patients between the 2 time periods, but the differences were not significant. An 8% decline in survival occurred among stage II patients, but this decline was also not significant (Table 7). When these cases were classified by stage and grade, a nonsignificant decreased survival rate was seen in the early stage, low grade group, with 5-year survival decreasing from 95% to 90% between the 2 time periods (Table 8). A nonsignificant decline in late stage, low grade tumors was also found, with survival decreasing from 73% to 63% from 1993-1996 to 1997-2000.
No significant change in survival was seen in ureteral carcinoma patients when classifying cases either by distinct stage (0a-IV) (Table 7) or by the categories of early versus late stage and low versus high grade tumors (Table 8). Although not significant, early stage tumors showed an increase in survival of 5% among low grade tumors and an increase in survival of 13% among high grade tumors between the 2 time periods.
A Cox proportional hazard model was used to assess the 5-year risk of death in patients diagnosed in 1997-2000 compared with those diagnosed in 1993-1996, adjusting for sex, age, race, stage, and grade. The model was run separately for early and late stage disease for each cancer site (data not shown). For late stage disease, the hazard ratio (HR) was slightly increased for bladder cancer in the later time period (HR, 1.04; 95% confidence interval [CI], 1.02-1.07). For late stage renal pelvis and ureter cancers, the hazard ratio was not significantly different between the 2 time periods. Among early stage cancers, there was a small but significant increase in the risk of death in the later time period (1997-2000) for bladder (HR, 1.07; 95% CI, 1.05-1.09) and renal pelvis cancers (HR, 1.16; 95% CI, 1.03-1.30). The hazard ratio for ureter showed a significant decrease in the risk of death in the later time period (HR, 0.86; 95% CI, 0.76-0.98).
Our data showed a small stage migration in renal pelvis and ureteral cancers from 1993 to 2005, with a slightly increased proportion of early stage tumors, due primarily to an increased proportion of stage 0a tumors and a decreased proportion of stage II tumors. No stage migration from early to late stage or vice versa was seen in bladder cancers from 1993 to 2005; however, within the category of early stage disease, the proportion of stage 0a cases increased, whereas the proportion of stage I cases decreased. The increase in early stage tumors is unclear, but may reflect an increased use of computerized tomography urography.17 This study has been reported to have a sensitivity in the 90% range for detecting upper tract malignancies, better than that of standard computed tomography scans.18
A limited number of other studies have investigated stage migration patterns in TCC of the bladder and upper urinary tract. Vaidya et al analyzed 176 patients with bladder TCC who underwent cystectomy between 1992 and 1999.19 At their initial presentation, 57% of patients had de novo muscle-invasive disease. These data were compared with older series from the early 1980s by Kaye and Lange20 and Hopkins et al,21 which found rates of de novo muscle-invasive disease of 84% and 91%, respectively. Despite inherent pitfalls in comparing the data of Vaidya et al to the earlier studies, a change in stage at presentation over time seemed likely.
Nieder et al analyzed trends in bladder cancer presentation and survival among Florida patients diagnosed between 1981 and 2004.22 Among the 87,978 cases, the stage at diagnosis did not change significantly. In situ (Ta and Tis) cancer incidence increased by 1% per year; local disease (T1, T2a, T2b) incidence decreased by 1% annually; and advanced disease (T3a, T3b, T4, nodal and distant metastases) incidence decreased by 0.74%. When data were analyzed by sex and ethnicity, the decrease in advanced cases was accounted for mainly by the change in incidence among men of all ethnicities.
By using the Surveillance, Epidemiology, and End Results (SEER) database, Munoz and Ellison found 9072 incident cases of upper urinary tract TCC, including 5394 cases of renal pelvis TCC and 3678 cases of ureteral TCC.3 A comparison of stage at presentation between 1973-1984 and 1985-1996 found the rate of in situ disease increased from 7.2% to 23%; the rate of local disease decreased from 50.4% to 31.8%; the rate of regional disease changed from 33.8% to 37.1%; and the rate of distant disease went from 8.5% to 8.0% (P < .001 for all stages).
The benefit of bladder cancer screening programs, which could theoretically uncover the condition at an earlier stage, remains unclear. A retrospective study by Messing et al comparing men whose bladder cancer was diagnosed after participating in at-home screening with men diagnosed with bladder TCC without prior screening revealed much higher rates of muscle invasion in the nonscreened patients (122 of 511 patients, or 23.9%) than in the screened group (1 of 21 patients or 4.8%).23 If further studies confirm this benefit, routine screening of high-risk populations could possibly lead to diagnosis at a lower stage.24
Our data show that in a comparison of diagnosis years 1993-1996 and 1997-2000, 5-year survival rates for all stages of bladder, ureteral, and renal pelvis TCC—with the exception of stage I bladder TCC—did not significantly change. Survival rates in stage I bladder decreased slightly, and stage I and II renal pelvis showed nonsignificant declines in survival. After adjusting for stage and grade, only renal pelvis late stage, low grade tumors showed a slight decline in survival, suggesting that the increase in the proportion of high grade tumors accounts for part of the decline in survival. Cox regression results also showed a small increase in risk of death in 1997-2000 for early stage bladder/renal pelvis tumors after adjusting for grade and other factors.
Several studies have shown that higher tumor grade is a risk factor for lower progression-free survival,25, 26 although these studies did not address overall survival. It is possible that the small decrease in 5-year survival for stage I bladder cancer could be because of the increase in the proportion of high grade tumors, although other factors could contribute, as well. The cause of the increased proportion of high grade tumors is not entirely clear. Grading of these tumors has been the same throughout all editions of the ICD-O manuals, so a change in ICD-O grading criteria over time cannot be implicated, despite the change in grading criteria of the separate WHO/International Society of Urological Pathology consensus classification that occurred during our study period.27, 28 In addition, we found that the increase in high grade tumors was seen in the patient population with a first or single primary tumor; therefore, the trend cannot be attributed solely to those patients with a history of multiple tumors.
Munoz and Ellison identified 5-year disease-specific survival rates for upper tract TCC patients in the SEER program between 1973 and 1996.3 Among those with renal pelvis TCC, the 5-year survival rate for in situ disease was 95.7%, for localized disease 87.8%, for regional disease 62.4%, and for distant disease 14.6%. Among those with ureteral TCC, the 5-year disease-specific survival rate was 94.5% for in situ disease, 90.5% for localized disease, 62.8% for regional disease, and 21.8% for distant disease. (These survival rates are from before some of the diagnosis dates [1997-2000] for this study, however.) Among 252 patients with upper tract TCC treated surgically at the University of Texas Southwestern Medical Center between 1960 and 1992, Hall et al identified survival rates of 100% for Ta/cis, 92% for T1 disease, 73% for T2 disease, and 41% for T3 disease. The median survival for patients with T4 disease was 6 months.29
The lack of substantial improvement in survival rates for bladder, renal pelvis, and ureteral tumors may also be due in part to the lack of new therapies for TCC introduced during this time period. Several trials investigating perioperative chemotherapy showed a benefit of neoadjuvant chemotherapy in the late 1990s and early part of this decade,30 but it is unlikely that perioperative chemotherapy was in widespread use during our study period. There have also been few improvements in overall survival with therapies for metastatic TCC. The combination of methotrexate, vinblastine, doxorubicin, and cisplatin (MVAC) had been the standard chemotherapy regimen since its development in the 1980s31-33 until von der Maase et al demonstrated that the combination of gemcitabine and cisplatin (GC) had a better toxicity and tolerability profile.34 Median survival, however, was not significantly different between the 2 regimens: 13.8 months with GC versus 14.8 months with MVAC. Nephrouretectomy has been the standard therapy for localized upper tract TCC. Adjuvant chemotherapy has proven tolerable in locally advanced disease,30 but the rarity of the condition has made large prospective trials difficult to perform. In addition, suboptimal utilization of therapies that have been shown to improve survival may also contribute to a lack of survival improvement with time. For example, we have previously reported the low incidence of perioperative chemotherapy for patients in the NCDB with stage III bladder TCC diagnosed between 1998 and 2003.35 The increasing incidence of high grade tumors in the backdrop of current treatment paradigms may also have likely implications on survival. For example, Lambert et al found that whereas the use of intravesical therapy for high grade T1 bladder cancers increased between 1990 and 2005, disease-free survival for these patients decreased, possibly because of delays in what may be the optimal therapy for such patients—radical cystectomy.36 Why we detected a small increase in the risk of death for early stage bladder/renal pelvis tumors but not ureteral tumors after adjusting for grade between diagnosis years 1997-2000 compared with 1993-1996 is unclear, and will require further study as more survival data become available in the current decade.
The lack of a significant improvement in survival in urothelial tumors between 1993 and 2000 may also be because of the finding that survival rates include deaths from all causes. As urothelial malignancies are linked to smoking, and the majority of patients in our study are older than 60 years, the prevalence of other comorbidities, such as heart and lung disease, is likely high. In addition, the prevalence of such comorbid conditions in the study population could have increased with time, possibly because of factors such as the increasing proportion of patients older than 80 years. Although the relative survival rates used in this study are age adjusted, smoking associated comorbidities may be higher among those with urothelial tumors than in the general population. A lack of change in the survival from these conditions could very well have impacted the overall survival of the patients we studied.
Our study has several strengths. First, the National Cancer Database is 1 of the largest sources of patient-specific diagnostic and treatment information in the United States. The data have the specific AJCC T, N, and M components, allowing us to bridge data by diagnosis year and TNM edition when T, N, and M definitions and stage definitions changed over time for bladder cancers. The data are abstracted from the patient medical chart by trained cancer registrars.
However, there are several limitations. Selection bias may be present, because NCDB collects data only from CoC-approved facilities; however, it captures 75% of the to 18% of patients with missing data have a markedly different stage distribution than those not missing stage data. In addition, the lack of standardized and centralized pathologic review precluded a uniform diagnostic assessment of the malignancies—both of the diagnoses and of the tumor grade—and possible discrepancies between different pathologists' grade classifications could definitely affect the presented data.
Renal pelvis and ureteral cancers showed a slight increase in the proportion of early stage cases between 1993 and 2005. In bladder cancer, no change in the overall proportion of early stage cases was seen; however, an increase in the proportion of stage 0a cases was seen with a decrease in the proportion of stage I cases. The percentage of high grade tumors in all disease sites increased during the study period. Furthermore, 5-year overall survival rates did not significantly change during this time period, underscoring the need for continued advances in surgery and chemotherapy for the treatment of urothelial malignancies to improve patient outcomes.
Conflict of Interest Disclosures
Support provided by Empire Clinical Research Investigator Program, Project 10 (K.A.D.).
- 14International Classification of Diseases for Oncology. 3rd ed. Geneva, Switzerland: World Health Organization; 2000: 71-72.
- 15Commission on Cancer: Facility Oncology Registry Data Standards Revised for 2007. Chicago, IL: American College of Surgeons; 2007.
- 16Statistics and Epidemiology for Cancer Registrars, Self-instruction Manual for Cancer Registrars, Book 7. Surveillance Epidemiology and End Results Program. Bethesda, MD: National Institutes of Health; 1995., , , , , .
- 28Sternberg's Diagnostic Surgical Pathology. 4th ed. Philadelphia, PA: Lippincott, Williams and Wilkins; 2004..