Effects of tumor size and location on survival in upper tract urothelial carcinoma after nephroureterectomy.

INTRODUCTION: Upper Tract Urothelial Carcinoma (UTUC) is a rare disease with few prognostic determinants. We sought to evaluate the impact of tumor size and location on patient survival following nephroureterectomy for UTUC.
MATERIALS AND METHODS: Data on 8284 patients treated with radical nephroureterectomy for UTUC in the United States between 1998 and 2011 were analyzed from the National Cancer Data Base. Univariable survivorship curves were generated based on pT stage, pN stage, grade, tumor size, and tumor site (renal pelvis vs. ureter). A Cox proportional hazards model was used to evaluate the effect of age, comorbidity, T stage, lymph node involvement, tumor site, and tumor size on survival.
RESULTS: The median follow-up time was 46 months. A majority of the patients were male (55.4%) with a tumor size of ≥3.5 cm (52.0%) and pT stage <T2 (47.8%). The overall 5 years survival overall survival (OS) for the entire cohort was 51.6%. When stratified by tumor size <3.5 cm or ≥3.5 cm the 5-year OS was 45.9% and 58.5%, respectively. On multivariable analysis controlling for age, Charlson comorbidity index, grade, and tumor stage, tumor size ≥3.5 cm was independently predictive of worse OS (odds ratio: 1.13 [95% confidence interval: 1.02-1.26], P = 0.023).
CONCLUSIONS: Using the largest series of patients with UTUC undergoing nephroureterectomy, we demonstrated a worse survival in patients with larger tumor sizes (≥3.5 cm) but no difference in survival based on tumor location while controlling for other pathologic characteristics. Incorporation of tumor size into perioperative risk modeling may help with patient stratification and provide further prognostic information for patient counseling.

INTRODUCTION

Upper tract urothelial carcinoma (UTUC) is a rare disease representing 5% of all urothelial malignancies, with an incidence of 2.08 cases per 100,000 person years in the United States.[1] UTUC is generally an aggressive disease, and it is estimated that at diagnosis, 60% of UTUCs are invasive, compared to 15%–25% of urothelial carcinoma of the bladder.[2] Due to its rarity, clinical decision-making surrounding treatment of UTUC is largely extrapolated from the existing literature for urothelial carcinoma of the bladder.

To better understand the nature of UTUC, perioperative predictive models have been developed to predict invasive, and non-organ confined disease spread, recurrence-free survival (RFS), and cancer-specific survival (CSS).[345] While tumor stage and grade are consistently important in prognosis, tumor size and location have been less thoroughly studied, though recent retrospective studies have found that tumors >3 cm in size are associated with worse recurrence-free and CSS, and overall survival (OS).[678] Furthermore, there is controversy within the literature with regards to the impact of tumor location (renal pelvis vs. ureter) on survival outcomes. Whereas several studies of smaller cohorts appear to indicate that tumor location is predictive of postoperative outcomes,[9101112] other groups failed to show such a correlation.[131415] The analysis of the largest prospective cohort of UTUC patients to date (the UTUC Collaboration) did not find tumor location to be of prognostic import.[1617]

In this study, we utilized the National Cancer Data Base (NCDB) to better understand the impact of tumor size and tumor location on survival and to provide risk factor prognostic information for patients with UTUC. We hypothesize that larger tumor size and renal pelvis tumor location are associated with worse survival. Consideration of tumor size and location in addition to other patient and pathologic characteristics may allow for more precise risk stratification of patients with UTUC with the potential to identify patients who may benefit from more timely and aggressive surgery, as well as consideration of neoadjuvant chemotherapy (NAC).

MATERIALS AND METHODS

Data were obtained from the NCDB participant user file. The NCDB is a national database of oncologic outcomes for more than 1500 Commission on Cancer accredited facilities in the United States and Puerto Rico, representing 70% of all cancer cases in the United States. De-identified individual patient data are linked with demographic, pathologic and survivorship data for analysis.

Inclusion criteria

We identified 8,284 patients treated for UTUC in the United States between 1998 and 2011 with localized disease (cN0/cNx, cM0) for urothelial carcinoma (ICD-0-3 codes: 8120/3) with the location in the renal pelvis or ureter (C65-C68). All patients were treated with radical nephroureterectomy (RNU), with or without bladder cuff excision, and had no prior malignancy. Bladder cuff excision was not recorded; hence, the cohort includes a mix of these surgical variants.

Patient demographic data

Cohort demographic variables included age, sex, race/ethnicity, Charlson comorbidity index (CCI), location of treatment by region, year of treatment, and treatment center type. Race/Ethnicity data were categorized as white/black/other/unknown. The confidence interval was divided into groups as no comorbidities, CCI = 1, and CCI >1. Regions of the treatment within the United States consisted of Northeast, South/Southeast, Midwest, and West. Years of treatment were stratified into 1998–2000, 2001–2003, 2004–2006, and 2007–2011. Treatment center types included community cancer programs, comprehensive community programs, academic/research program, and other. Community cancer and comprehensive community programs differ in the number of cases they treat, at 100–500 new cases to over 500 cases, respectively. Academic/research programs treat over 500 new cases per year and also train resident physicians in at least four areas.

Pathologic data

Primary clinical variables included tumor size (<3.5 cm or ≥3.5 cm), tumor grade (low/high), tumor stage (pT0/T1/T2/T3/T4), pN stage (pN0/N+, Nx), and primary site (renal pelvis, ureter). Tumor size of 3.5 cm was selected as this was the median tumor size in the group.

Statistical analysis

Primary outcome variables included OS and 5-year survival after nephroureterectomy for UTUC. Univariable survival analysis was performed using the Kaplan Meier method and compared using the log rank test. Multivariable survival analysis was performed using a Cox proportional hazards model to evaluate the effect of age, comorbidity, T stage, lymph node involvement, tumor site, tumor grade, and tumor size on 5-year OS. Statistical analysis was conducting using STATA software, version 9.0 (Stata Corporation, College Station, TX, USA). Statistical significance was set a priori at P < 0.05.

RESULTS

Basic demographics

A total of 8284 patients met inclusion criteria with median follow-up time of 46 mo. The incidence of UTUC increased with age with 56.7% of patients being older than 70 years of age. The cohort was skewed toward the male gender (55.3% male, 44.6% female). One-third (32.6%) of patients had no comorbidities based on CCI. Treatment location was well distributed among regions within the United States with a majority of patients being treated at academic and comprehensive community programs. The number of patients with tumor size < 3.5 cm was almost equal to the number of those with tumor size ≥ 3.5 cm (48.0% vs. 52.0%) with an interquartile range of (2.3–5.0 cm). The majority of patients had high-grade tumors (82.0%), pN0 (68.3%), and a tumor location in the renal pelvis (68.0%). A larger proportion of patients presented with pT3 disease (31.2%) than pT0/Ta/TIS (24.2%) or pT1 (23.6%) disease [Table 1].

Table 1

Patient and demographic characteristics for patients with 5-year survival data

Survival analysis

The 5-year survival for the entire cohort was 51.6%. The 5-year OS decreased with each successive increase in pT stage. The 5-year survival was worse for pN + versus pN0 (14.8% vs. 55.8%, P < 0.0001), high grade tumors versus low grade tumors (48.4% vs. 74.3%, P < 0.0001), and tumor size ≥ 3.5 cm versus < 3.5 cm (45.9% vs. 58.5%, P < 0.0001). Having a tumor located in the ureter had similar 5-year OS as tumors in the renal pelvis (51.8% vs. 51.5%, P = 0.817) [Table 2]. Kaplan Meier survival curves are demonstrated in Figure 1 for pathologic characteristics.

Table 2

Univariate 5-year survival by pathologic characteristics

Figure 1

Survivorship (n = 8,284) by tumor size P Log Rank <0.0001 (a), N stage P Log Rank <0.0001 (b), grade P Log Rank <0.0001 (c), T stage P Log Rank <0.0001 (d), and tumor site P Log rank = 0.817 (e)

Multivariable survival analysis demonstrated that increasing age (P < 0.001), increasing comorbidity (P < 0.001), larger tumor size (P = 0.043), high grade tumors (P < 0.001), and increasing pT stage are associated with worse OS outcomes [Table 3]. Tumor location was not independently predictive of survival.

Table 3

Cox proportional hazards model for mortality risk at 5 years

DISCUSSION

The rarity of UTUC has made it difficult to ascertain the prognostic value of several important disease variables. The NCDB database contains the largest cohort of such patients to date, from which we were able to analyze survival outcomes from 8284 patients who underwent radical nephroureterectomy for UTUC. We demonstrate that while controlling for age, comorbidity, tumor grade, pT stage, and pN stage, tumor size ≥3.5 cm was associated with worse OS. Location of the tumor (renal pelvis vs. ureter) did not have an impact on survival.

The role of tumor size as a prognostic factor for survival after RNU has been controversial. Some studies showed that a tumor diameter of 3.0 cm or above is a risk factor for poor RFS after RNU, and Pieras et al. showed that tumor size can be used as a prognostic factor for bladder recurrence.[718] Conversely, others have shown that tumor size did not have an effect on CSS or recurrence.[619] The variation in findings is, in part, due to limitations in sample sizes. A larger cohort by Shibing et al. analyzing 795 patients from several centers did find that tumor size >3.0 cm was an independent predictor of worse RFS, CSS, and OS.[8]

Our study, which represents the largest such retrospective cohort, had a median tumor size of 3.5 cm, similar to the median of 3.0 cm found in other cohorts. As such, a cutoff of 3.5 cm was chosen. Our findings suggest that tumor size can be used as a predictor of 5-year survival, with rates of 58.5% for ≤3.5 cm versus 45.9% for those larger than 3.5 cm.

Another controversy in UTUC is whether tumor location (renal pelvis vs. ureter) affects survival and recurrence. Here again, the existing literature is conflicting. Two smaller single-institution studies showed that carcinoma of the proximal ureter has worse prognosis.[1120] Another study showed that ureteral tumors have worse RFS and CSS on multivariable analysis.[21] However, a large number of other studies show that tumor location does not affect survival outcomes after surgery.[1513141622] In the present study, we confirm that there is no statistically significant survival difference between patients with tumors primarily in the renal pelvis or in the ureter.

Identification of prognostic factors in UTUC is crucial to inform treatment strategies. The appropriate use of perioperative chemotherapy in UTUC might be better guided through an understanding of which patients are likely to have more aggressive disease characteristics. Adjuvant chemotherapy (AC) for UTUC, for example, has had varied success. Multiple studies have shown that AC has little or no effect on overall or CSS.[2324] Moreover, RNU itself may block patients from receiving AC; some studies have shown a 25%–30% reduction in patient eligibility due to loss of estimated glomerular filtration rate.[2425] For these reasons, NAC is becoming the gold standard for treatment of UTUC. And although prior evidence for the efficacy of NAC in UTUC came from extrapolating data demonstrating NAC effectiveness in bladder cancer, an important recent retrospective review by Porten et al. evaluating NAC for UTUC showed better overall and disease-specific survival for patients receiving NAC.[26] Based on our current understanding, it is difficult to know which patients will benefit most, if at all, from chemotherapy. Identifying pre- and post-operative clinicopathologic factors may, therefore, help to develop a risk-adapted approach to patient selection. Unfortunately, the use of perioperative chemotherapy was not widespread during the period in our study; and the patients in our NCDB cohort did not have sufficiently robust perioperative chemotherapy data to make conclusions.

This study has some important limitations. Retrospective data collection brings with it innate weaknesses. As with most retrospective clinical datasets, the data are affected by selection bias and confounding by indication to treat. Surgeons do not all use the same standardized criteria to decide whether or not to perform lymphadenectomy, and lymphadenectomy templates differ, all of which may impact survival outcomes. Indeed, there is biological plausibility that more aggressive lymph node dissection may improve survival following RNU, though no definitive evidence yet exists.[27] Furthermore, as UTUC is predominantly a disease of the elderly, competing comorbid illness likely plays an important role in the long-term outcomes, though this was controlled to the best of our ability using the comorbidity index. Finally, our dataset did not include measurements of lymphovascular invasion and tumor architecture, which have been found to be prognostic factors in other studies.[416]

CONCLUSIONS

Using the largest series of patients with UTUC having undergone RNU, we demonstrated worse survival in patients with larger tumor sizes (≥3.5 cm) and no difference in survival based on tumor location after controlling for other clinical and pathologic characteristics. Incorporation of tumor size into perioperative risk models may provide important prognostic information for patient counseling and selection regarding perioperative chemotherapy. We also confirm in this study that tumor location (renal pelvis vs. ureter) does not affect survival. Further prospective studies are needed to confirm the clinical and prognostic value of tumor size.